Optimization of the HEC-RAS Based Flood Inundation Mapping using Adaptive Neuro- Fuzzy Inference System: A case study of Olkeriai River Basin, Kenya

Optimization of the HEC-RAS Based Flood Inundation Mapping using Adaptive Neuro- Fuzzy Inference System: A case study of Olkeriai River Basin, Kenya | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Optimization of the HEC-RAS Based Flood Inundation Mapping using Adaptive Neuro- Fuzzy Inference System: A case study of Olkeriai River Basin, Kenya Enock Bore, Achieng Kevin Otieno, Duncan Maina Kimwatu, Jeremiah K. Kiptala This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6670542/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 05 Nov, 2025 Read the published version in Modeling Earth Systems and Environment → Version 1 posted 12 You are reading this latest preprint version Abstract Effective flood modelling is essential in flood disasters’ impact reduction and sustainable land use planning, particularly in vulnerable areas such as the Olkeriai River Basin in Kenya. This study provides an innovative hybrid model of Adaptive Neuro-Fuzzy Inference System (ANFIS) and the Hydrologic Engineering Center-River Analysis System (HEC-RAS) model for improved spatial accuracy in flood inundation mapping. The coupled model provides flood inundation mapping for the whole catchment area unlike HEC-RAS model which is restricted to the defined river bank lines. Flooding in the Olkeriai River basin continues to disrupt riparian agriculture and settlements in this basin, but most conventional hydrological models tend to not accurately simulate flood extents over varied terrain. The steady flow simulation in HEC-RAS was used to simulate a 100-yr return period flood with peak flows from a calibrated hydrologic model in Hydrologic Engineering Center- Hydrologic Modelling System (HEC-HMS). Historical events of flooding and conditioning factors were used to train ANFIS model to create a spatial flood Inundation index map. Lastly, HEC-RAS flood depth inundation outputs were calibrated by overlaying them on the flood inundation index map based on ANFIS model outputs. Results indicate that ANFIS model worked well in terms of accuracy and prediction (R² = 0.960, RMSE = 0.092, MAE = 0.090 and AOC = 0.910), and hybrid model enhanced flood prediction capability (R²= 0.944, RMSE = 0.445, MAE = 0.337 and NSE = 0.944). Derived flood inundation map delineates the high-risk areas within and outside the river corridor. These outcomes will enable local authorities, disaster managers, and planners to implement effective actions in flood mitigation, plan early warnings, and assist land-use planning that renders the community more resilient. Flood modelling flood resilience HEC-RAS ANFIS HEC-HMS hybrid model flood hazard map Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Figure 12 Figure 13 Figure 14 Introduction Flooding refers to an overflow of water onto the land (Douben, 2006 ). The Federal Emergency Management Agency (FEMA) gives a precise definition of a flood as a general and temporary condition of partial or complete inundation of normally dry areas of land or property (Iii, 2005 ). Likewise, the Intergovernmental Panel on Climate Change (IPCC) in its Special report on managing the risks of extreme events and disasters to advance climate change adaptation classifies floods as the overflowing of the usual limits of a stream or other watercourse, or the accumulation of water over areas which are not usually covered by water (Ahmed et al, 2007 ). All these definitions by the authoritative bodies refer to the main idea of water going beyond its normal limits and spilling over into regions that are normally dry as per (Nordlo, 2016 ). According to Gaume et al. ( 2016 ), the two broad categories of flooding are general flooding and flash flooding. While both are characterized by an overflow of water, both are significantly different in the temporal character they display (Now, 2018 ). General flooding is generally a longer-term phenomenon that will likely last days or weeks (Munyai et al., 2021 ). Conversely, a flash flood is defined by its sudden occurrence, typically brought about by an abnormal or heavy amount of rain in a brief time, typically in under six hours (Feng et al., 2021 ). Flash floods are commonly described as turbulent water torrents that can easily devastate riverbeds, urban streets, or mountain canyons and carry everything with them (Khan et al., 2013 ). From the general phenomenon of floods, flood occurrences are usually classified according to their main causes and the surrounding environments in which they take place according to a study by Eskresi, ( 2015 ). Two main categories are fluvial flooding and alluvial flooding (Hamerlynck et al., 2013 ; Eskresi, 2015 ; Mensah, 2020 ). Riverine flooding, also called fluvial flooding, happens when water in a stream, river, or other watercourse rises and spills over the banks and into the surrounding low-lying areas, which are the natural floodplains (Bernhofen et al., 2018 ). This kind of flooding is most often due to the overflow of heavy or prolonged rain that exceeds the natural capacity of the river channel (Bernhofen et al., 2018 ). Fluvial floodings may happen as overbank flooding, with the water level barely exceeding above the stream or river banks, or flash flooding in channels of an already established riverbed from a massive amount of water surging through with little notice (Merz et al., 2010 ). Alluvial flooding, however, is a unique category of flooding occurring on alluvial fans or the same landforms (Piracha, 2021 ). Such fan-shaped landforms are developed at the apex, usually where a mountain creek with steep slope meets a relatively flat valley floor. Alluvial flooding involves high-velocity flows, active erosion, sediment transport, and deposition processes, and also irregular flow paths (Piracha, 2021 ). Such flooding occurs frequently at the foot of mountain ranges in semi-arid and arid lands (Risi et al., 2018 ). Debris flow, which occurs as an intense, concentrated, high-speed water flow carrying a high proportion of sediment and debris, and debris flood with a moderate sediment concentration, may be associated with alluvial fan flooding (Ing, 2010). The processes may move immense amounts of material and are one of the principal sources of hazards (Risi et al., 2018 ). The causes of flooding are varied and function at all geographical scales, ranging from global climatic processes to local environmental and anthropogenic conditions (Mfon et al., 2022 ). Understanding these causes is necessary to successfully develop flood inundation and susceptibility maps. Globally, climate change is a leading cause of growing flood risk (Rannie, 2016 ; Wahyudi, 2020 ). Global warming produces more intense precipitation events, in that, the warmer atmosphere is able to hold more water, resulting in heavier and more frequent rainfall (Ogega, 2018 ). Sea level rise heightens coastal flooding risks (Hallegatte et al., 2022 ). Mfon et al. ( 2022 ) noted that urbanization of areas at flood risk further aggravates risk by contributing more runoff from impervious surfaces. Looking to regional causative factors, a case study by Mensah ( 2020 ) demonstrates coupled causative factors for flood-causing situations. The study notes that heavy rain is one such regular cause and that the El Niño climatic phenomenon has also caused devastating flooding in East Africa. Poor planning and infrastructure are the primary causes of flood disasters in Ghana and Nigeria (Mensah, 2020 ). Locally, the direct causes of flooding are excessive rain beyond the capacity of the ground to absorb it, overflows of rivers and lakes, sudden snowmelt, coastal storm surges, failure of dams or levees, and river ice jams (Masese et al., 2016 ). Flood risk is also increased locally by urbanization through impervious surfaces and plugged drainage systems (Njogu, 2021 ). The world is facing an increasing threat from flooding, a natural hazard that has become more frequent and intense in recent decades (Risa, 2016 ; Mfon et al., 2022 ; Maranzoni et al., 2023 ). Floods are the deadliest natural hazards, striking numerous regions in the world each year (Berthelot et al., 2014 ). Between the years 1994 and 2013, floods affected nearly 2.5 billion people worldwide, causing more than $ 40 billion in damage each year (CRED, 2015 ). Also, in the past three decades, the average number of flood-related disasters has increased by nearly 35% (Nordlo, 2016 ). Between 1990–2022, floods affected more than 3 billion people worldwide with 218,353 deaths reported (World, 2024 ). Between 80–90% of all documented disasters from natural hazards during the past 10 years have resulted from floods, droughts, tropical cyclones, heat waves and severe storms (Piracha, 2021 ). In Africa, the period between 1950 and 2019, small and medium scale floods affected millions of people across the continent affecting nearly seven million people and caused 27,000 deaths (Tramblay et al., 2020 ). The prevalence rates of floods in Kenya stands at 27% and takes 5% of the population affected by disasters as reported by Okaka ( 2019 ) and the most commonly affected places by floods are the floodplains of the major rivers such as the lower Tana River, the lower Nzoia River in Budalang’i and also the emerging alluvial river systems in arid and semi-arid regions (Njogu, 2021 ) According to Morris ( 2014 ), the effects of flooding are deep, reaching the natural and human environments in deep and profound ways. The study also notes that the effects can be long-lasting and devastating. From an ecological perspective, flooding devastates habitats in wetlands and forest ecosystems as it leads to soil and riverbank erosion and sedimentation that affect aquatic organisms. Of specific concern is sewage, agricultural runoff, and industrial chemical water pollution, which poses a health risk. Freshwater aquatic organisms may also be affected by flooding. Small floods have beneficial impacts such as recharging groundwater and sedimenting nutrient-dense sediment behind them. Studies by Munyai et al. ( 2021 ) and Risa ( 2016 ) in Zimbabwe and Nigeria respectively indicated that flooding has detrimental impacts on plant, soil stability, and ecosystem process. The socio-economic damage caused by floods is also catastrophic. CRED ( 2015 ) analyses the impacts of disasters in that flooding causes enormous infrastructural and property loss and thereby enormous economic damage. The study also notes that entrepreneurial ventures are halted, and agricultural activities are slowed down by destroying crops and animals, people get displaced as an aftermath, waterborne and vector-borne illnesses and psychological effects are the health repercussions. According to Maranzoni et al. ( 2023 ), flood inundation mapping relies on a variety of modelling approaches, which can in general be distinguished along a spectrum from physically-based models that reproduce the underlying physical processes to data-driven models that apply statistical and machine learning algorithms to detect patterns and make forecasts. Tshimanga et al. ( 2016 ) also notes that knowledge of the nature of these various types of models is needed to select the most suitable method to an application in flood inundation mapping. Hydrologic Engineering Center-River Analysis System (HEC-RAS), one of the flood simulation tools, is a popular software package for hydraulic computation in river systems (Costabile et al., 2020 ). It applies hydraulic principles to simulate water flow and interaction with natural and human made features. HEC-RAS is applied in floodplain mapping, flood insurance studies, bridge and culvert design, dam break analysis, and channel modification studies (Umamahesh, 2019). It can also be applied for real-time flood forecasting in combination with hydrologic models such as Hydrologic Engineering Center- Hydrologic Modelling System (HEC-HMS). Nordlo (2017) notes that, HEC-RAS has benefits which include the capability to simulate sophisticated hydraulic processes, accommodating a range of flow regimes such as hydraulic structures, ease of use with a Graphic User Interface (GUI) and free availability and that the newer versions of HEC-RAS from version 5.0 to version 6.2 accommodate 2D modeling capabilities. According to Costabile et al. ( 2020 ), limitations of HEC-RAS include possible numerical instability, requirement of large geometric data, limitation of 1D models to simulate complex flow, requirement of careful calibration and validation, very large computational time for 2D models, and sensitivity to the quality of input data. HEC-HMS is a computer program hydrologic simulation employed to model precipitation-runoff processes in watershed systems (USACE, 1998 ). HEC-HMS models parameters such as precipitation, evapotranspiration, snowmelt, infiltration, runoff, and channel flow as per the HEC-HMS manual USACE, ( 1998 ). The HEC-HMS manual also notes that, this hydrologic model computes flood discharges and produces flow hydrographs that are generally used as an input to hydraulic models such as HEC-RAS. It is employed for flood frequency analysis and also for simulating the effect of land use change on runoff as per the manual USACE, ( 1998 ). The strengths noted from the HEC-HMS manual include large watershed simulation, capability to simulate urban and natural systems, capability to employ many modelling methodologies, integrated GUI, continuous simulation support, and free distribution. Also, the weaknesses noted from the manual are lack of explicit model for hydraulic channel flow complexities, dependency on external GIS packages, complex calibration, and possible limitations in highly steep channels or systems with considerable backwater effects (Guo et al., 2021 ). Studies which include Jonkman et al. ( 2008 ), Setiawan et al. ( 2018 ) have demonstrated the application of HEC-HMS with HEC-RAS and GIS in flood modelling and hazard mapping within defined river bank lines. Adaptive Neuro-Fuzzy Inference System (ANFIS) is a hybrid artificial intelligence that combines neural networks and fuzzy logic to model complex, nonlinear relationships from data (Walia, 2015 ). ANFIS is data-trained to tailor the parameters of a fuzzy system, generally a Takagi-Sugeno model (Walia, 2015 ). ANFIS has been extensively used in flood susceptibility mapping to estimate flood probability in terms of topography, meteorology, and land use (Ullah, 2013 ). ANFIS is also applied to river flow rate and water level flood forecasting (Al-hmouz et al., 2012 ). Walia ( 2015 ) notes the benefits and limitations associated with ANFIS, the benefits noted include the ability to learn nonlinear relationships, integration of learning and reasoning, flexibility, dealing with uncertainty, and relatively fast learning rates. Limitations noted are based on training data quality dependency, curse of dimensionality as a result of a high number of input variables, optimal parameters being difficult to select, absence of apparent physical sense, danger of getting stuck in local minima, and variation of performance depending on architecture and training algorithm (Bui et al., 2018 ). Studies like Al-hmouz et al. ( 2012 ) and Bui et al. ( 2018 ) have established the efficiency of ANFIS in flood susceptibility mapping and forecasting in different regions. Additionally, geospatial technology is useful in flood control and mapping flood risk of hazard areas, i.e., it can be applied prior to floods in areas prone to floods in order to prepare cases of future events, during floods to track floods and post-floods for damage evaluation and countermeasure measures (Saudi, 2023 ). In the present study, GIS integrated with hydraulic model, HEC-RAS are used together with remote sensing to simulate results for flood inundation extent. The main objective was to model flooding extent of Olkeriai River basin for the steady flow analysis by optimizing the outputs of HEC-RAS model using Adaptive Neuro-Fuzzy Inference System (ANFIS) machine learning algorithm. The objective was to improve the spatial precision and predictive reliability of flood mapping in this extreme condition so that land-use planning and disaster preparedness are formulated according to the worst-case but feasible scenario. The need to model for steady flow conditions is the reason why HEC-RAS was chosen for this study as this model is particularly strong is such conditions therefore allow more accurate representation of flood events as noted by Costabile et al. ( 2020 ) which also shows that HEC-RAS has the ability to generate flood plain maps and flood extent maps with high spatial accuracy. However, HEC-RAS uses limited data inputs which includes river discharge data, land use land cover data, digital elevation model of the area and the soil characteristics of the study area which limits the model’s capacity to account for broader spatial variability and interactions across the floodplain (Al, 2017 ). It is also noted by the same study that HEC-RAS has a limitation of confined simulation within the defined bank lines, its spatial limitation to the river corridor presents a knowledge gap in landscape-scale flood hazard mapping. These shortcomings limit the comprehensive mapping of flood zones in the area. By incorporating multiple floods contributing factors which influence flooding beyond the river corridor as input variables for the model, ANFIS machine learning algorithm was therefore incorporated for this study to optimize the outputs obtained from HEC-RAS and to accurately predict and to extend flood inundation mapping coverage beyond the riverbanks, to map flood extents across the catchment area. This improves the accuracy of flood inundation mapping by taking advantage of ANFIS capability in detecting intricate relationships among model inputs and flood extents observed. This method brings a paradigm change in flood hazard mapping that further helps in formulating flood mitigation strategies. Flood hazard awareness and avoidance will improve the resilience of smallholder farming families by safeguarding their crops and livelihood from damages caused by floods. Methodology Description of the study area The Olkeriai basin is located in Kajiado county in Kenya and lies between longitude 37⁰25’00”E to 30⁰ 50’00”E and latitude 1⁰45’00”S to 2⁰30’00”S with an approximate area of 1524 Km 2 (Fig. 1 ). The river is seasonal and sandy with a length of approximately 100 km and forms one of the tributaries that drain its water into river Athi basin. The basin, is semi-arid with bimodal rainfall, with short rains occurring between October and December and long rains from March to May, ranging from 300 to 800 mm annually. The temperature ranges from high temperatures of 28°C to 35°C and low temperatures of 17°C to 22°C, with January and February being the hottest months (Uker et al., 2022 ). 60 km stretch of the river where agriculture is mainly practiced and experiences riparian flooding near Mashuuru town was used for analysis and modelling. The basin also features land with semi-arid savannah plains intermixed with seasonal river valley areas (Matsaba et al., 2021 ). Geology in this region includes deposits from volcanic processes of Quaternary and Pleistocene together with basement rocks, which in turn produce types of soil as black cotton soil and sandy loams (Uker et al., 2022 ). This region is socio-economically dependent on pastoralism and plough agriculture dependent on the semi-arid climatic conditions coupled with unpredictable regimes of rainfall (Duker et al., 2022 ). Datasets used for flood inundation mapping using HEC-HMS, HEC-RAS and ANFIS The study utilised several datasets as presented in Table 1 , which are, European Space Agency (ESA) Sentinel-2 (10-meter 2024 imagery) for curve number/Manning's coefficients and Land Use/Land Cover (LULC) classification, Global Hydrologic Soil Groups (A-D classification) from Distributed Active Archive Center - Global Hydrologic Soil Groups (DAAC-GHSOGS) for Modelling in HEC-HMS for estimation of runoff, Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) daily data as input in HEC-HMS meteorological component, European Space Agency (ESA) 30- metre spatial resolution Copernicus Digital Elevation Model (DEM) for mapping terrain and flow in the HEC-HMS and HEC-RAS models, and Water Resources Authority (WRA) discharge data for flood modelling in the HEC-HMS model and calibration. Elevation, slope, lineament density and Topographical Wetness Index (TWI) were extracted from a 30m spatial resolution Copernicus DEM and used as inputs for the ANFIS model development. The LULC, Normalized Difference Vegetation Index (NDVI) data maps and Manning’s Coefficient which are also the input datasets for the ANFIS model development were created by carrying out supervised classification of a 10m spatial resolution Sentinel-2image acquired using Google Earth Engine platform from Copernicus. Table 1 Datasets used, their description and their relevance in the study. Dataset Source Specification Relevance Sentinel-2 European Space Agency (ESA) Multispectral 2024 imagery, 10m resolution LULC classification to derive CN & Manning’s roughness coefficients. Also used to derived NDVI which is an input data for ANFIS model development. Soils(GHSOGS) Global Hydrologic Soil Groups Distributed Active Archive Center (DAAC) Hydrologic soil group classification (A-D), global coverage, based on soil texture& infiltration rates Essential for creating the curve number (CN) map in HEC-HMS to estimate runoff potential based on soil moisture retention in the basin. Rainfall CHIRPS (Climate Hazards Group InfraRed Precipitation with Station data) Daily rainfall data Provides meteorological input for HEC-HMS to simulate runoff and discharge. DEM Copernicus Programme (European Space Agency) 30m resolution digital elevation model, derived from satellite altimetry Used for terrain mapping, watershed delineation, flow direction, and stream network generation in HEC-HMS and HEC-RAS to model hydrologic and hydraulic responses. Also used to extract elevation, slope, lineament density and TWI which are the input data for ANFIS model development. Discharge Water Resources Authority (WRA) Observed discharge Calibrates Observed discharge which serves as input in HEC-RAS model. Also, an input in ANFIS model development. Workflow for flood inundation mapping using HEC-HMS, HEC-RAS and ANFIS The flowchart in Fig. 2 shows the methodology for flood inundation mapping in Olkeriai River Basin with hydrological and hydraulic modelling using HEC-HMS and HEC-RAS respectively, ANFIS model development and the process of optimization of HEC-RAS outputs. The process was carried out first, through terrain delineation and watershed setup to create subbasins and reaches in the basin. Supervised pixel based random trees classification was then carried out to produce LULC map which was used together with soils data to create curve number map. These were then utilized as inputs to HEC-HMS for basin model configuration with addition of lag time. The HEC-HMS model was then calibrated with the observed Olkeriai River basin flow data for the March 2024 flooding event against daily discharge at Mashuuru to generate daily discharge and annual maximum data for Olkeriai basin. With annual maximum discharges obtained by the HEC-HMS model, Flood Frequency Analysis (FFA) was then done in the basin to determine peak discharges for the 100-year flood return period for the basin. Steady flow analysis was then simulated using HEC-RAS to obtain flood velocity and depth maps of the basin flood plains using the steady discharge from the 100-year return period. The outputs from HEC-RAS together with the modelled daily discharge data, elevation data, TWI data, slope, lineament density, soil data, LULC data for the year 2024 and the NDVI data and the HEC-RAS output to train and validate ANFIS model. Optimization process was achieved by overlaying the HEC-RAS outputs on the flood inundation index map based on ANFIS machine learning algorithm followed by extraction of corresponding pixel values in HEC-RAS and ANFIS outputs then an optimized flood depth inundation map produced. The flow discharge data for the March 2024 floods for Olkeriai River basin was used to calibrate and validate the HEC-HMS for this study. Steady flow analysis was utilized in this study because of its applicability in simulation under peak flood conditions where the flow conditions are considered to be constant with respect to time (Shabir, 2016 ). Steady flow analysis is a simple but effective method for flood depth and area determination in worst flood conditions (Access, 2019 ). 100-year return period flood was employed since it is a severe flood with 1% probability of occurrence in any year, according to Conventional Engineering and Risk Management Practice for the Estimation of Flood Hazards. The 100-year flood extent results were optimized using ANFIS machine learning algorithm, as it is the situation that poses maximum threat to life, property, and agriculture in the Olkeriai River Basin. Modelling Olkeriai River discharge data using HEC-HMS HEC-HMS was first used to simulate Olkeriai river basin flow discharge data for the steady flow analysis with HEC-RAS. The model was coupled with a 30 years period 1994–2024 daily rainfall dataset from CHIRPS with a Curve Number (CN) map derived from global hydrological soil groups (DAAC) and land use land cover map, lag time estimates, and topographical data obtained from Copernicus 30m DEM. The Olkeriai River basin’s model was delineated by marking the watershed and defining its physical characteristics with the help of Copernicus 30m DEM. The DEM provided high-resolution data regarding flow direction, accumulation, and stream networks for the accurate representation of the hydrologic response of the basin. A total of 15 subbasins were generated and one was merged to as they were too small in area. The CN map developed using global hydrological soil groups and LULC map, created according to the Soil Conservation Service (SCS) method to estimate runoff potential, was averaged for each sub basin through zonal statistics. Lags, defined as the time difference between peak rainfall and peak discharge, were estimated by empirical relationships using the watershed lag method by United States Department of Agriculture (USDA), Natural Resources Conservation Service (NRCS) (USDA and NRCS 2008), as a function of basin characteristics using Eqs. 1 and 2 as in Table 2 . $$\:L=0.6{T}_{c}$$ 1 \(\:{T}_{c}=\:\frac{{\mathcal{l}}^{0.8}{(S+1)}^{0.7}}{1140{Y}^{0.5}}\) (2) where: L is lag, h; \(\:{T}_{c}\) is the time of concentration, t; λ = flow length, ft; Y is the average watershed land slope, percent; S is the maximum potential retention obtained using Eq. 3 $$\:S=\:\frac{1000}{\text{c}\text{n}{\prime\:}}-10$$ 3 where: cn′ is the retardance factor which is approximately the same as the curve number (CN). Table 2 Curve Number, Slope, Time of Concentration and Lag of each subbasin in Olkeriai River Basin Subbasin CN Slope T c Lag (min) Subbasin-1 87.76 1.39 2.62 94.21 Subbasin-3 89.67 1.15 0.95 34.09 Subbasin-4 87.46 1.43 3.20 115.14 Subbasin-5 86.89 1.51 2.96 106.38 Subbasin-6 87.36 1.45 1.00 36.07 Subbasin-7 87.58 1.42 1.96 70.74 Subbasin-8 86.66 1.54 3.12 112.40 Subbasin-9 87.36 1.45 1.40 50.48 Subbasin-10 88.52 1.30 1.77 63.76 Subbasin-11 88.06 1.36 2.27 81.64 Subbasin-12 87.50 1.43 1.62 58.16 Subbasin-13 88.75 1.27 1.33 48.00 Subbasin-14 91.62 0.91 0.97 34.79 Subbasin-15 87.38 1.44 3.61 129.85 The Soil Conservation Service Curve Number (SCS-CN) was chosen as the loss method in the hydrologic modelling because it could accommodate the soil moisture and land use characteristics. This method of estimating runoff involves rainfall, soil type, and land cover with less input parameters required curve number and initial abstraction. The SCS-CN procedure in HEC-HMS estimates the runoff based on assigning CN values to LULC and HSGs with Olkeriai basin having soil groups C, C/D, D, and D/D. CN values were assigned based on the HEC-HMS manual USACE ( 1998 ), these CN values varied between 30 and 100, with higher numbers representing more runoff as shown in Table 3 . After preparation of the curve number map using the LULC and the hydrological soil groups, the average curve number for each subbasin was calculated using the zonal statistics tool in ArcGIS pro software. Table 3 Table for Assigning SCS Curve Numbers for HEC-HMS (Source: USACE ( 1998 )) LULC Type HSG C HSG C/D HSG D Agriculture 79 85.0 91 Sand 74 75.5 77 Developed 88 91.0 94 Bare land 84 87.5 91 Thick Forest 77 81.5 88 Sparse Forest 85 88.0 91 Shrubland 81 86.5 92 Wetlands 99 92.5 100 The transform method used was the Snyder Unit Hydrograph method. This method uses parameters such as basin lag time and peak rate factor, (USACE, 1998 ), which were calibrated using observed flow data. For routing, the Muskingum method as described by Cunderlik, ( 2004 ) was applied to simulate water movement through river channels, balancing accuracy and computational efficiency. A meteorological component was developed from the 30-year time series rainfall data obtained from CHIRPS and treated to account for the spatial and temporal distribution within the study area. Hydrologic simulation in HEC-HMS was performed considering steady flow conditions. Seventy percent of the observed discharge data for the March 2024 flooding event in the Olkeriai River basin was used to calibrate the model, also, manual calibration was done by adjustment of parameters such as SCS unit hydrograph values, lag times, graph type, and Muskingum x and k coefficients (Costabile et al., 2020 ). Univariate Gradient optimization simulation was created to further improve the accuracy of the model through automated calibration (Cunderlik, 2004 ). The remaining 30% of the discharge data was used to validate the model. Flood frequency analysis (FFA) for steady flow simulations The process of steady flow analysis was done by first designing steady peak discharge related to floods that have 100-year probability of occurrences. This is useful in modelling the worst-case scenario floods that are extremely severe in case of occurrence. With annual maximum discharges obtained by the HEC-HMS model, Flood Frequency Analysis (FFA) was done in the basin to determine peak discharge of this flood return period. The log-persons type III probability distribution was used in calculation of discharges corresponding to a given probability of exceedance and it has been clearly defined and implied by Mendez et al. ( 2023 ). The probability of exceedance is defined as the inverse of the return period as per Head et al. ( 2023 ) for example the 100-year return period corresponds to a 1% probability of exceedance which translates to 1% chance of happening at any year, calculated as in Eq. 4 . $$\:P=\:\frac{1}{T}$$ 4 Where P is the probability and T is the return period. This method has been applied successful in flood frequency analysis where it has been used in Kenya by Lang’at et al. (2019) in Tana River basin and Houessou-dossou et al. ( 2023 ) for FFA analysis in Narok town. Steady flow simulation using HEC-RAS In steady flow simulation using HEC-RAS, the 30m Copernicus DEM was used to generate terrain data. Then the river center line, bank lines, flow paths and cross sections was manually generated. Then manning roughness coefficients were assigned to the river bed and bank lines based on land use land cover. A steady data profile was created where discharge from FFA were used for the 100-year return period profile. Steady flow analysis was simulated to obtain flood extent and depth maps of the basin flood plains. For result validation, the produced flood extent maps were compared to field observations and historical flood extent information gathered by interviewing local people during a field survey carried out in the region. Development of Adaptive Neuro-Fuzzy Inference System (ANFIS) model Adaptive Neuro-Fuzzy Inference System (ANFIS) was implemented to model flood inundation index map within the Olkeriai River Basin based on HEC-RAS model outputs and geospatial flood conditioning factors. The model incorporated the result of the 100-year flood simulation by HEC-RAS and the flood causative factors to generate a spatially continuous flood depth inundation map, which was utilized in the optimization process. Flood depth output from HEC-RAS steady-flow simulation for return period of 100 years was used as the dependent variable in ANFIS model. HEC-RAS flood depth output was exported as a raster layer and resampled to spatial resolution of 30 meters, matching Landsat-derived NDVI and DEM resolution. The Universal Transverse Mercator (UTM), Zone 37S, World Geodetic System 1984 (WGS 84) datum was used as the base coordinate reference system for all the spatial layers in order to have consistency and spatial integrity. Six independent variables (flooding conditioning factors) were used because of their established relevance in flood inundation studies in the literature, they include the following; elevation, slope, mean annual rainfall, stream flow, NDVI, TWI. All raster layers were reprojected, clipped to the area of study, resampled at 30 m resolution, and co-registered through the use of the SNAP toolbox in ArcGIS pro software. To standardize the inputs to be fed into the ANFIS model, min-max normalization was applied on every continuous input, converting the values into normalized range of 0 to 1 using the formulae shown in Eq. 5 . This normalization method provided evenly weighted scaling to all the inputs so that the neural-fuzzy model is able to converge efficiently during training. This method has already been applied and shown successful in flood modelling by (Palakondi, 2017). $$\:{y}_{n}=\frac{y-{y}_{1}}{{y}_{2}-y}$$ 5 Where; \(\:{y}_{n}\) is normalized value (transformed within 0and 1 range), \(\:y\) the original influencing factor value at the sampling point, \(\:{y}_{1}\) is the minimum value of a factor across all locations and, \(\:{y}_{2}\) represents the maximum value of a factor across locations respectively. For training the ANFIS model, a sample dataset was created by extracting values from each of the input raster layers and the corresponding flood depth raster. 10,000 spatial sample points were generated within the study area using a stratified random sampling process within ArcGIS, with such, a representative coverage was obtained across different landform and flood-prone areas. For every sample point, the input pixel values of the six input variables were extracted, as well as the corresponding flood depth value as the output target. The created dataset (10,000 records × 7 variables) was exported in Comma-Separated values (CSV) format and imported as an input to the ANFIS model in Python. The data set was divided into 70% for training (7,000 records) and 30% for testing (3,000 records). The training data set was utilized to create the neuro-fuzzy inference rules, whereas the test data set was reserved for model evaluation and validation. ANFIS model development was executed using Python code of local code in the available open-source platforms created using NumPy and TensorFlow. A Gaussian membership function was applied to all the input variables to be utilized in the fuzzy set representation, and the system employed a first-order Sugeno-type fuzzy inference system (Walia, 2015 ). The number of membership functions in every input was manually tuned to obtain the best complexity and accuracy level. Hybrid optimization algorithm with least-squares estimation for linear parameters and backpropagation in order to optimize nonlinear parameters was employed until convergence and stability of the validation error was attained using the procedure of Walia ( 2015 ). Following successful training, the ANFIS model was executed using the whole raster stack (6 layers) over the study area. The trained model was input with all pixels' values in order to calculate a flood inundation index score, generating a continuous raster surface between 0 (low inundation) and 1 (high inundation). The performance of the model was evaluated using the test dataset where calculated flood inundation scores were compared against the observed depth of flood. Model performance metrics which were used are; Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Coefficient of Determination (R²). Furthermore, the flood inundation map was cross-validated with the Receiver Operating Characteristic (ROC) curve, and observed flood areas (from the 100-year simulation) were used as reference. The Area Under the Curve (AUC) was estimated, where AUC > 0.80 was taken as an indication of high spatial prediction performance following a criterion by Janssens, ( 2020 ). Optimizing HEC-RAS Simulation Using ANFIS Data to develop a hybrid model The sample datasets were obtained by overlaying HEC-RAS flood inundation map on the developed flood inundation index map based on ANFIS then extracting corresponding pixel values that are matching using the spatial analyst extension tool in ArcGIS pro software. HEC-RAS data provided outputs that are restricted along the river geometry which include variables between flood depth and velocity maps. The flood inundation index map based on ANFIS provided a broader output coverage since it extends to cover the whole catchment area through its ability to integrate multiple environmental and hydrological factors in its prediction. The specific flood depth values from HEC-RAS were matched to correspond with values from ANFIS flood inundation index map. In dealing with erroneous or missing data values, mean substitution for numerical data and interpolation techniques for other data types was used. The Z-score statistical method with 95% confidence level was used to detect outliers for removal because these points had the potential to distort the model's performance results. Data cleaning was done to minimize data distortion caused by missing or incorrect entries. Stratified random sampling method in ArcGIS was used for dataset selection to pick data points evenly across all conditions in the river catchment from the HEC-RAS and ANFIS datasets. Three hundred data points were effectively collected across all depths of floodwaters along with various types of terrain and hydrological contexts essential for building an effective model. The datasets were divided into two distinct sets, one of which trained the model and a second sample set to test model performance. Training dataset consisted of 70% of three hundred sample points, this provided the model with the ability to know how to make model predictions based on HEC-RAS flood simulation. Test part of the model comprised 30% of data to test its performance in coping with unknown data. Stratified sampling was employed to ensure data distribution in both subsets. Geographically Weighted Regression (GWR) was utilized to correlate HEC-RAS flood depth results to spatial ANFIS model outputs. The model was trained on 70% of the data and was implemented by choosing an optimal kernel and bandwidth by using cross-validation. Bias correction was done to overcome the HEC-RAS model restriction that operated only within defined river geometry. The evaluation for identifying systematic biases in data proceeded through the utilization of Bland-Altman Analysis as per Fernandez ( 2009 ), that measured differences in the HEC-RAS values ANFIS developed map. The deviations of the mean values from zero indicates that HEC-RAS outputs had regular deviations from ANFIS values either in positive or negative direction. Quantile Mapping and Bias Adjustment Factors were applied in analysis for bias correction process. Quantile Mapping as stated Hamill ( 2018 ) was employed as a technique of matching the cumulative distribution functions (CDF) of outputs from HEC-RAS and ANFIS data. The corrected HEC-RAS outputs were tuned with this method by aligning them with the statistical requirements of the ANFIS dataset. Bias Adjustment Factor was calculated by checking systematic differences among the original datasets. Bias Adjustment Factor allowed the removal of residual biases from HEC-RAS output data through its application. For validation of the model, 30% of the randomly picked data points was used to validate calibrated and bias-corrected model. The metrics which was used to evaluate the performance of the optimized model included the Coefficient of Determination (R 2 ), Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Nash- Sutcliffe Efficiency (NSE) and the mean bias error. To improve model performance, the analysis process was repeated several times while fine tuning the hyperparameters and optimization procedures. Final Optimized HEC-RAS based Flood Inundation Simulation The process involved utilizing the trained and optimized model for HEC-RAS output simulation across the entire river catchment region. The model used ANFIS output data to predict water depth extent and hydraulic conditions throughout the whole catchment. The simulated output from HEC-RAS optimization covered the entire area of the river catchment solving the limitation of HEC-RAS of being confined within the defined river geometry. Results Simulated river flow discharge using HEC-HMS Model On running the HEC-HMS hydrological simulation, a Root Mean Square Error (RMSE) standard deviation of 0.411 and a Nash-Sutcliffe Efficiency (NSE) of 0.848 was obtained after final calibration of the model and can be illustrated in Fig. 3. This enabled the simulation of the 30 years rainfall to obtain credible discharges that are crucial in modelling flood inundation extents for the study. The HEC-HMS model successfully modelled daily discharges from the year 1994 to 2024 as shown in Fig. 4 in order to identify extreme hazard flooding events in the Olkeriai basin and for steady flow modelling in HEC-RAS. The model results showed close corelations between extreme rainfall and high discharges that lead to flooding of the river as depicted in Fig. 4. Maximum annual discharge data was also modelled as shown in Fig. 5 for use in flood frequency analysis. Flood Frequency Analysis was then performed using the annual discharge data shown in Fig. 5. This was to determine peak discharge for 100- year flood return period for the basin which was to be used for steady flow analysis in HEC-RAS. The determined peak discharge is illustrated by the curve in Fig. 6. The peak discharge obtained for the 100-year return period was 3749.3 m 3 /s. The discharge data was now ready for modelling of flood extents in the HEC-RAS model. Simulated flood inundations for the 100-year flood using HEC-RAS Model Steady flow analysis was simulated to obtain flood inundation extent and depth maps of the basin flood plains using the steady data profile which was created where discharge from FFA were used for the 100-year; 3749.3 m3/s return period profile. The 100-year flood extent map shown in Fig. 7 flood covers 37.27 km² with an average depth of 3.22 meters. From the river corridor, small water extent coverage is seen on the Northern region of the study area which is a high elevated region while large extents are observed on the lower region of the study area. Higher water depth values are also observed on the southern region of the study area and along the river channel indicated by deep green colour coding. This agrees with the observation for the water inundation coverage that a larger extent is observed on the Southern region of the map, this implies that the southern region of the map is more susceptible to flooding due to its low elevation. Low depth values are observed mainly in the northern region of the study area. Flood inundations index mapping using Adaptive Neuro-Fuzzy Inference System (ANFIS) To evaluate the performance of the trained ANFIS model, observed flood depths (from 100-year simulation with HEC-RAS) were compared to ANFIS-predicted values. The following statistical measures below were calculated: Root Mean Square Error (RMSE) of 0.092, Mean Absolute Error (MAE) of 0.090, and Coefficient of Determination (R²) of 0.960 represented in Table 4. Low RMSE values indicates that the model performed well with low prediction errors. These results represent good model performance, with small error values and a high R² score representing the model capability to explain flood depth variability accurately. The performance of the model in classification accuracy was also tested through a Receiver Operating Characteristic (ROC) curve. This was achieved through the development of predicted flood inundation index scores against flood inundation depth values from the 100-year flood inundation map. As shown in Table 4, the derived Area Under the Curve (AUC) was 0.910. AUC value more than 0.90 indicates that the model strongly discriminates flood-prone and non-flood-prone regions over the study area. Table 4 ANFIS model performance metrics and their corresponding values Model performance metrics Metric value Root Mean Square Error 0.092 Mean Absolute Error 0.090 Coefficient of Determination 0.961 Area Under Curve 0.910 The trained ANFIS model was then used spatially to produce a continuous flood inundation index map of the Olkeriai River Basin. The flood causative factor values of each pixel were used as inputs to the trained model to calculate the flood inundation indices between 0 (very low) and 1 (very high). The resulting flood inundation index map is presented in Fig 8. Higher depth values are seen along the river geometry on the central region of the study area. Optimized HEC-RAS based Flood Inundation using ANFIS data The hybrid model of ANFIS and HEC-RAS achieved an NSE value of 0.944 together with Coefficient of Determination, R 2 value of 0.944. This indicates 94.4% of the observed data variance thereby demonstrating excellent reliability and prediction abilities as shown in Fig. 9. Fig. 10 is graphical representation showing model predictions of water depth between the HEC-RAS model along with the ANFIS-calibrated version against validation points while displaying varying degrees of uncertainty intervals at 68% and 95% and 99.7%. The calibrated ANFIS model appears as the blue line alongside the original HEC-RAS model that shows its outputs as the orange line. The alignment between the two curves reveals strong agreement especially in the shaded uncertainty areas. Based on results in Fig. 11, Fig. 12 and Table 5 , the hybrid model shows high reliability in its validation stages. The developed predictive model produces Coefficient of Determination value of 0.944, Root Mean Square Error (RMSE) values of 0.445 and Mean Absolute Error (MAE) of 0.337 which illustrate its predictive power through low error estimation. Validation process used a total of 275 observation points from a sample size of 300 observed data points, 25 data points were removed since they were outliers and could bring skewedness in the prediction. Results from the Mean Bias Error evaluation show 0.098 as the calculated value indicating no major distortion that would impact the model performance. The overall model validation results prove that the model is well-optimized, accurate, and highly suitable for prediction. Table 5 Summary of model validation metrics and their corresponding values for the optimized model Model validation metric Metric value Root Mean Square Error 0.445 Mean Absolute Error 0.337 Coefficient of Determination 0.944 Nash-Sutcliffe Efficiency 0.944 Mean Bias Error 0.098 Fig. 13, optimized HEC-RAS flood depth extent map shows predicted depth values in the entire river catchment by spatial distribution ranging from 0.09m to 10.09m. Low depth values showed regions which have low inundations and lower risk of flooding and mostly appear on the upper region of the basin that has higher elevation. Higher depth values on the other hand that indicate regions that are at greater risk of flooding within the basin occupying most of the region along the river Olkeriai corridor and on the regions with low elevation. The predicted values were reclassified into three categories based on Jenks method of flood hazard risk categorization as shown by Indexes (2021) and then visualized in GIS as shown in Fig. 14 . The gradient color scheme presents prediction intensity variations through its green-to-red color transition. Estimated values that are red or orange appear mostly in river corridors and lower elevation zones thus marking potential vulnerable flood areas and high-flow zones. The green sections denote regions with reduced predicted values along with the upland areas and areas less prone to flooding. Table 6 shows the area extent of the reclassified map based on the inundation indices. The highest probability of flooding exists in moderate-risk zones occupying 41.09% of the catchment area mostly located in low elevated areas in the northern region of the study area. These areas need strict flood mitigation through barriers and improved drainage systems along with built structure controls. High-risk zones which account for 39.23% of the total area which occupies the area along the river corridor and low elevated zones in the southern part of the area shows occasional yet seasonal flood patterns during extreme weather conditions. The area of the basin amounting to 19.68% occupied low-risk region as these locations either rest at elevated heights or possess good drainage characteristics mostly seen in the western and North western region of the area. Table 6 Statistics showing area extent in percentage of reclassified optimized flood risk map. Flood risk category Pixel count Percentage cover High risk zone 669331 39.23 Moderate risk zone 700911 41.09 Low risk zone 335639 19.68 Discussions The HEC-RAS simulated flood and optimized hybrid ANFIS-HEC-RAS model clearly illustrates the flood behaviour of the Olkeriai River basin. HEC-RAS simulated a 100-year return period flood which covered an area of 37.27 km² with an average depth of 3.22 meters. Small extent coverage from the river corridor is seen on the upper side of the study area with high elevation while large coverage of water is seen on the southern region of the study area due to the low elevation of this region, water pooling being common in such regions causing flooding. Based on these results, immediate measures are therefore needed to curb the dangers of floods in case of any occurrence mostly on the Sothern part of Olkeriai River basin. According to a study by Mcmahon ( 2025 ) on recognition and implications of high- depth and energy flood events in Scotland, similar results as these indicate a high-magnitude, high-energy flood event capable of significantly impacting riparian food production in agriculture and proximal settlements. Similar results were reported by Tamiru ( 2022 ) on using machine learning and HECRAS integrated models for flood inundation mapping in Baro River Basin, where extreme-event, maximum depth foods were large and perilous to cropland and rural infrastructure. The study indicated that floods in alluvial low-lying floodplains result in extensive flooding and deposition that reduce soil and agricultural yields' quality. The spatial water depth and inundation distribution of the HEC-RAS results (Fig. 7 ) are a result of local land use, channel morphological geometry, and local topography. This corroborates with an explanation by Malarchick ( 2019 ), describing the hydrogeomorphic and geomorphic response to extreme flood events and how channel confinements and elevation gradient dictate floodwater routing and energies of flow over the duration of extreme events. Though HEC-RAS-enabled useful channel-based predictions of flooding are made simpler, its model is only applied to the defined bank lines ignoring the overbank flow processes or flood extension onto the larger catchment. The HEC-RAS-optimized model, extends flood inundation mapping and forecasting to the whole catchment area using spatially distributed data and learning nonlinear relationships from HEC-RAS output, (Fig. 7 ). This was achieved by incorporating multiple floods contributing factors which influence flooding beyond the river corridor. The strong consistency of HEC-RAS and ANFIS-predicted water depths within overlapping areas confirms that the hybrid method effectively surpassed the physical accuracy of the HEC-RAS model while extending its spatial applicability. The optimized output achieved an extremely low RMSE of 0.445 and a virtually zero bias of 0.098, testifying to the strength of the ANFIS-based improvement. These values validate the ability of ANFIS to generalize complex hydrological interactions, particularly under ungauged or data-scarce conditions. This is also evidenced by the work of Bui et al. ( 2018 ), who used ANFIS to model flood behaviour in the Haraz watershed, Iran and obtained better prediction than that from conventional HEC-HMS model. Likewise, Ullah ( 2013 ) highlighted that coupling machine learning with physical models minimizes structural and data-related uncertainties and results in more effective and spatially representative flood simulations. In this research, ANFIS enabled flood inundation mapping throughout the entire catchment, beyond river geometry bounded by HEC-RAS, hence facilitating more integrated flood risk assessment. The reclassified results from an optimized model showed regions with different risks of flooding (Table 6 ). High risk flooding zones accounted for 39.23% of the total area, occupied most of the area along the river corridor and low elevated zones in the southern part of the area. Moderate risk zones took occupying 41.09% of the catchment area mostly located in low elevated areas in the northern region of the study area while the low-risk zones took the least percentage with only 19.68%. These results call for immediate measures in most of the parts of the study area to mitigate the impending danger of floods in case of occurrence. Results of this research further attest to the emerging trend of integrating physically based hydraulic models with data-driven models like ANFIS to enhance accuracy and spatial coverage in flood forecasting. To a catchment that is likely to flood with susceptible agriculture and infrastructure, applying these hybrid methods in such a setting has tangible added value in relevant advice for the local planners, policymakers, and emergency managers. Effective maps of floods can direct zoning control, warning infrastructure, and site-based schemes of mitigation like levees or detention ponds to improve community resilience. Conclusion This research illustrates that it is possible to achieve much better accuracy and spatial coverage of flood prediction by integrating traditional hydraulic modeling and machine learning algorithms. The novelty is that although HEC-RAS can produce a good simulation of the flood in the river channel, it does not have capacity to simulate the wider floodplain beyond the defined river bank lines. By incorporating ANFIS model which is more data-driven, the shortcomings of the HEC-RAS model were overcome to create a hybrid model with an ability to express the depth as well as the spatial distribution of the flood extremely reliably. The optimized model not only improves predictive performance but also allows for more realistic, evidence-based mapping of flood risk throughout the whole catchment. This method allows for more informed decision-making in flood management, land use planning, and disaster preparedness in at-risk river basins. In general, the results support the potential of integrating physically based models with artificial intelligence to develop more comprehensive and beneficial flood risk assessments. Declarations No any conflict of interest was declared throughout the research process. Acknowledgements I wish to acknowledge my supervisors for their support throughout the research process. I also wish to thank the data providers; Water Resource Authority, European Space Agency (ESA), USGS Earth Explorer, Distributed Active Archive Center (DAAC), Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS). Also, I acknowledge the authors I cited in this study for providing vital information for conducting this research work Authors’ contributions Funding The research was funded by the Dutch Research Council (NWO) and the Directorate-General of International Cooperation (DGIS) of the Netherlands Ministry of Foreign Affairs for DUPC3 (2021-2027) Water and Development Partnership Programme of the Smallholder farming families Adapt African Alluvial Aquifers to Strengthen Their Own Resilience (A4Store) Project. Availability of data and materials The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. Ethics approval All authors have read, understood, and have complied as applicable with the statement on “Ethical responsibilities of Authors” as found in the Instructions for Authors. Competing interests The authors declare no competing interests. References Access, O. (2019). The comparative study of flood modelling with the unsteady and the steady flow on Ngotok river The comparative study of flood modelling with the unsteady and the steady flow on Ngotok river . https://doi.org/10.1088/1757-899X/669/1/012018 Ahmed et al. (2007). Summary for policy- makers, climate change, IPCC WG1 fourth assessment report. Cambridge Universtity Press . http://ipcc.ch/meetings/session17/doc3d.pdf%5Cnhttp://link.springer.com/10.1007/BF02986817 Al-hmouz, A., Shen, J., Member, S., Al-hmouz, R., & Yan, J. (2012). Modeling and Simulation of an Adaptive Neuro-Fuzzy Inference System ( ANFIS ) for Mobile Learning. IEEE Transactions on Learning Technologies , 5 (3), 226–237. https://doi.org/10.1109/TLT.2011.36 Al, N. et. (2017). Potentials and limitations of 1D, 2D and coupled 1D-2D flood modelling in HEC-RAS. Tvvr17/5003 , 8. Ashok, V., & Umamahesh, R. N. V. (2019). Assessment of inundation risk in urban floods using HEC RAS 2D . 0123456789 . Bernhofen, M. V, Whyman, C., Trigg, M. A., Sleigh, P. A., & Smith, A. M. (2018). A first collective validation of global fluvial flood models for major floods in Nigeria and A fi rst collective validation of global fl uvial fl ood models for major fl oods in Nigeria and Mozambique . Berthelot, J., Saint-laurent, D., Gervais-beaulac, V., & Savoie, D. (2014). Assessing the Effects of Periodic Flooding on the Population Structure and Recruitment Rates of Riparian Tree Forests . 2614–2633. https://doi.org/10.3390/w6092614 Bui, D. T., Khosravi, K., Li, S., Shahabi, H., Panahi, M., & Ahmad, B. Bin. (2018). New Hybrids of ANFIS with Several Optimization Algorithms for Flood Susceptibility Modeling . https://doi.org/10.3390/w10091210 Costabile, P., Costanzo, C., Ferraro, D., Macchione, F., & Petaccia, G. (2020). Performances of the new HEC-RAS version 5 for 2-D hydrodynamic-based rainfall-runoff simulations at basin scale: Comparison with a state-of-the art model. Water (Switzerland) , 12 (9), 1–19. https://doi.org/10.3390/W12092326 CRED. (2015). The Human Cost of Natural Disasters. Centre for Research on the Epidemiology of Disasters CRED , 58. Cunderlik. (2004). Calibration, Verification and Sensitivity analysis of the HEC-HMS hydrologic model. Journal of Chemical Information and Modeling , 53 (9), 1689–1699. Douben, K. (2006). CHARACTERISTICS OF RIVER FLOODS AND FLOODING : A GLOBAL . 21 (April), 25–27. Duker, A. E. C., Karimba, B. M., Wani, G. E., & Prasad, P. (2022). Security in flexibility : Accessing land and water for irrigation in Kenya-s changing rural environment Security in fl exibility : accessing land and water for irrigation in Kenya ’ s changing rural environment . March . https://doi.org/10.1051/cagri/2022003 Eskresi, P. (2015). Coping with Floods . July . https://doi.org/10.1007/978-94-011-1098-3 Feng, B., Zhang, Y., & Bourke, R. (2021). Urbanization impacts on flood risks based on urban growth data and coupled flood models. Natural Hazards , 106 (1), 613–627. https://doi.org/10.1007/s11069-020-04480-0 Fernandez. (2009). Validating the Bland-Altman Method of Agreement . 1–17. French, J., & Ing, R. O. Y. (2010). Mortality from Flash Floods : a Review of National Weather Service Reports , 1969-81 . July 1981 . Gaume, E., Borga, M., Llassat, M. C., Maouche, S., Gaume, E., Borga, M., Llassat, M. C., Maouche, S., & Lang, M. (2016). Mediterranean extreme floods and flash floods To cite this version : The Mediterranean Region under Climate Change. A Scientific Update , 133–144. Guo, K., Guan, M., & Yu, D. (2021). Urban surface water flood modelling – a comprehensive review of current models and future challenges . 2843–2860. Hallegatte, S., Ranger, N., Mestre, O., Corfee-morlot, P. D. J., Herweijer, C., & Muir, R. (2022). Assessing climate change impacts , sea level rise and storm surge risk in port cities : a case study on Copenhagen . 113–137. https://doi.org/10.1007/s10584-010-9978-3 Hamerlynck, O., Paul, J., Leauthaud, C., Cochet, H., Nyunja, J., Albergel, J., & Gru, O. (2013). Floods and livelihoods : The impact of changing water resources on wetland agro-ecological production systems in the Tana River Delta , Kenya . 23 , 252–263. Hamill. (2018). Probabilistic precipitation forecast postprocessing using quantile mapping and rank-weighted best-member dressing. Monthly Weather Review , 146 (12), 4079–4098. https://doi.org/10.1175/MWR-D-18-0147.1 Head, R., The, R., Exceedance, A., & Model, P. (2023). Revitalizing the Annual Exceedance Probability model (AEP) for use in the changing world Kaury Pena California State University Maritime Academy . 1–35. Houessou-dossou, E. A. Y., Gathenya, J. M., Njuguna, M., & Gariy, Z. A. (2023). Flood Frequency Analysis Using Participatory GIS and Rainfall Data for Two Stations in Narok . 1–12. Iii. (2005). federal register . Indexes, V. (2021). Effects of Different Normalization, Aggregation, and Classification Methods on the Construction of Flood Vulnerability Indexes . Janssens. (2020). Reflection on modern methods: Revisiting the area under the ROC Curve. International Journal of Epidemiology , 49 (4), 1397–1403. https://doi.org/10.1093/ije/dyz274 Jonkman, S. N., Bočkarjova, M., Kok, M., & Bernardini, P. (2008). Integrated hydrodynamic and economic modelling of flood damage in the Netherlands. Ecological Economics , 66 (1), 77–90. https://doi.org/10.1016/j.ecolecon.2007.12.022 Khan, A., Khan, B., Qasim, S., & Khan, S. (2013). Causes, Effects and Remedies: A Case Study of Rural Flooding in District Charsadda, Pakistan. Journal of Managerial … . Lang’at, Kumar, K. (2019). Identification of the Most Suitable Probability and Mean Streamflow . 1–24. Malarchick. (2019). ScholarWorks at University of Montana Philosophies of assertiveness Let us know how access to this document benefits you . Maranzoni, A., Oria, M. D., & Rizzo, C. (2023). Quantitative flood hazard assessment methods : A review . March 2022 , 1–31. https://doi.org/10.1111/jfr3.12855 Masese, A., Neyole, E., & Ombachi, N. (2016). Loss and Damage from Flooding In Lower Nyando Basin , Kisumu County , Kenya . September . Matsaba, E. O., van Selm, M., Wesonga, J. M., Goosen, H., Coninx, I., & Koomen, I. (2021). Agricultural Climate Atlas for Kajiado and Kiambu Counties, Kenya BT - Handbook of Climate Change Management: Research, Leadership, Transformation (J. M. Luetz & D. Ayal (eds.); pp. 3–23). Springer International Publishing. https://doi.org/10.1007/978-3-030-57281-5_227 Mcmahon. (2025). High-energy flood events recorded in the Mesoproterozoic Meall Dearg Formation , NW Scotland ; their recognition and implications for the study of pre-vegetation alluvium . 175 , 13–32. Mendez, M., Rica, D. C., Rica, C., José, S., & Rica, C. (2023). A comparison of generalized extreme value , gumbel , and log-pearson distributions for the development of intensity duration frequency curves . A case study in Costa Rica . 01002 , 1–8. Mensah. (2020). Causes , impacts and coping strategies of floods in Ghana : a systematic review. SN Applied Sciences , 2 (5), 1–13. https://doi.org/10.1007/s42452-020-2548-z Merz, B., Hall, J., Disse, M., & Schumann, A. (2010). Fluvial flood risk management in a changing world. Natural Hazards and Earth System Science , 10 (3), 509–527. https://doi.org/10.5194/nhess-10-509-2010 Mfon, I. E., Oguike, M. C., & Eteng, S. U. (2022). Causes and Effects of Flooding in Nigeria : A Review . 1 (9), 1777–1792. Morris. (2014). The impact of seasonal flooding on agriculture : the spring 2012 floods in Somerset , England . 7 (April 2012), 128–140. https://doi.org/10.1111/jfr3.12041 Munyai, R. B., Chikoore, H., Musyoki, A., Chakwizira, J., Muofhe, T. P., Xulu, N. G., & Manyanya, T. C. (2021). Vulnerability and Adaptation to Flood Hazards in Rural Settlements of Limpopo Province , South Africa . Njogu, H. W. (2021). Effects of floods on infrastructure users in Kenya. Journal of Flood Risk Management , 14 (4), 1–10. https://doi.org/10.1111/jfr3.12746 Nordlo. (2016). A review on mechanism of flood disaster management in Asia. International Review of Management and Marketing , 6 (1), 29–52. Now, A. (2018). Global Flooding . 8200 (2017), 47–52. https://doi.org/10.1080/19428200.2017.1390365 Ogega, O. M. (2018). Globalization and Global Warming: A Case of Laikipia County, Kenya. International Conference on Climate Change & Sustainable Development in Africa 2018 , 1 (Special Issue: Environment, Technology and Sustainable Development), 60–67. https://doi.org/10.26796/jenrm.v0i0.106 Okaka. (2019). Health vulnerability to flood-induced risks of households in flood-prone informal settlements in the Coastal City of Mombasa, Kenya. Natural Hazards , 99 (2), 1007–1029. https://doi.org/10.1007/s11069-019-03792-0 Palakonda, V., & Mallipeddi, R. (2017). Pareto Dominance-Based Algorithms With Ranking Methods for Many-Objective Optimization. IEEE Access , 5 , 11043–11053. https://doi.org/10.1109/ACCESS.2017.2716779 Piracha. (2021). Natural Disasters — Origins , Impacts , Management . 1101–1131. Rannie, W. (2016). The 1997 flood event in the Red River basin: Causes, assessment and damages. Canadian Water Resources Journal , 41 (1–2), 45–55. https://doi.org/10.1080/07011784.2015.1004198 Risa. (2016). Flooding as an environmental hazard . Risi, R. De, Jalayer, F., Paola, F. De, & Lindley, S. (2018). Delineation of flooding risk hotspots based on digital elevation model , calculated and historical flooding extents : the case of Ouagadougou. Stochastic Environmental Research and Risk Assessment , 32 (6), 1545–1559. https://doi.org/10.1007/s00477-017-1450-8 Saudi. (2023). Mapping of Groundwater, Flood, and Drought Potential Zones in Neom, Saudi Arabia, Using GIS and Remote Sensing Techniques . Setiawan, M. N., Kristanto, D., & Setiawan, J. (2018). Estimating design flood and HEC-RAS modelling approach for flood analysis in Bojonegoro city Estimating design flood and HEC-RAS modelling approach for flood analysis in Bojonegoro city . https://doi.org/10.1088/1757-899X/316/1/012042 Shabir. (2016). One Dimensional Steady Flow Analysis Using HECRAS – A case of River Jhelum, Jammu and Kashmir. European Scientific Journal, ESJ , 12 (32), 340. https://doi.org/10.19044/esj.2016.v12n32p340 Tamiru. (2022). Machine ‑ learning and HEC ‑ RAS integrated models for flood inundation mapping in Baro River Basin , Ethiopia. Modeling Earth Systems and Environment , 8 (2), 2291–2303. https://doi.org/10.1007/s40808-021-01175-8 Tramblay, Y., Villarini, G., & Zhang, W. (2020). Observed changes in flood hazard in Africa. Environmental Research Letters , 15 (10). https://doi.org/10.1088/1748-9326/abb90b Tshimanga, R. M., Marie, T. J., Kabuya, P., & Alsdorf, D. (2016). Flood forecasting : an international perspective A Regional Perceptive of Flood Forecasting and Disaster Management Systems for the Congo River Basin . April 2017 . Uker, A. E. C., Karimba, B. M., Wani, G. E., Prasad, P., Zaag, P. Vd. Der, & Fraiture, C. De. (2022). Security in fl exibility : accessing land and water for irrigation in Kenya ’ s changing rural environment . Ullah, N. (2013). Flood Flow Modeling in a River System Using Adaptive Neuro-Fuzzy Inference System . 2 (2), 54–68. https://doi.org/10.5296/emsd.v2i2.3738 USACE. (1998). HEC-HMS Technical Reference Manual Introduction. Computer Manual , 1–288. Wahyudi. (2020). The analysis of the causes of flood disasters and their impacts in the perspective of environmental law. IOP Conference Series: Earth and Environmental Science , 437 (1). https://doi.org/10.1088/1755-1315/437/1/012056 Walia, N. (2015). ANFIS : Adaptive Neuro-Fuzzy Inference System- A Survey . 123 (13), 32–38. World. (2024). Global , regional and national trends and impacts of natural floods , . 38 , 410–420. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 05 Nov, 2025 Read the published version in Modeling Earth Systems and Environment → Version 1 posted Editorial decision: Revision requested 29 Sep, 2025 Reviews received at journal 07 Sep, 2025 Reviews received at journal 20 Jun, 2025 Reviewers agreed at journal 31 May, 2025 Reviewers agreed at journal 29 May, 2025 Reviewers agreed at journal 28 May, 2025 Reviewers agreed at journal 28 May, 2025 Reviewers agreed at journal 28 May, 2025 Reviewers invited by journal 28 May, 2025 Editor assigned by journal 26 May, 2025 Submission checks completed at journal 26 May, 2025 First submitted to journal 15 May, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6670542","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":463244949,"identity":"0dff087a-a233-484f-a3bd-ff05bb843abc","order_by":0,"name":"Enock Bore","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA1UlEQVRIiWNgGAWjYBACAxhDgr35AIiSIUELz7EEMEWCFokcMJuwFnP23ocPPu6wkZfsOfP51Y0aCx4G9sNHN+DTYtlz3Nhw5pk0w9nsvdusc44BHcaTlnYDr8NupLFJ87YdTpDjObvNOIcNqEWCx4yQFvbff0FaJHKeGef8I04LGzMjUIu0RA7z49w2IrRY9hxjluxtSzOc2XPMjDm3T4KHjZBfzNnbGD/8bLORlzje/Phzzrc6OX72w8fwakEGbBJgkljlIMD8gRTVo2AUjIJRMHIAAGc/RT5M87cDAAAAAElFTkSuQmCC","orcid":"","institution":"Dedan Kimathi University of Technology","correspondingAuthor":true,"prefix":"","firstName":"Enock","middleName":"","lastName":"Bore","suffix":""},{"id":463244951,"identity":"d975a602-d5f1-4177-a352-dd2bb093b869","order_by":1,"name":"Achieng Kevin Otieno","email":"","orcid":"","institution":"Dedan Kimathi University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Achieng","middleName":"Kevin","lastName":"Otieno","suffix":""},{"id":463244953,"identity":"525fd09b-b100-415c-a5b6-1d2255028858","order_by":2,"name":"Duncan Maina Kimwatu","email":"","orcid":"","institution":"Dedan Kimathi University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Duncan","middleName":"Maina","lastName":"Kimwatu","suffix":""},{"id":463244956,"identity":"804595aa-2b83-48ab-910d-5032249324c8","order_by":3,"name":"Jeremiah K. Kiptala","email":"","orcid":"","institution":"Jomo Kenyatta University of Agriculture and Technology","correspondingAuthor":false,"prefix":"","firstName":"Jeremiah","middleName":"K.","lastName":"Kiptala","suffix":""}],"badges":[],"createdAt":"2025-05-15 08:38:22","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6670542/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6670542/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s40808-025-02667-7","type":"published","date":"2025-11-05T15:57:24+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":83611099,"identity":"3e99629b-bf01-4db4-b316-f8e02568f6ee","added_by":"auto","created_at":"2025-05-29 12:20:33","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":255257,"visible":true,"origin":"","legend":"\u003cp\u003eLocation of the Olkeriai sand river basin in Kenya.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6670542/v1/9ec4a3ea96052fbfae32550f.png"},{"id":83611100,"identity":"66f958dc-e63f-48a2-aa07-ea9ae90a6fd1","added_by":"auto","created_at":"2025-05-29 12:20:33","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":423523,"visible":true,"origin":"","legend":"\u003cp\u003eStudy methodology flow chart\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6670542/v1/7c0bede5ef7cea009b67dbe2.png"},{"id":83611963,"identity":"a1224448-71e8-4fef-b567-65ea3aba35b7","added_by":"auto","created_at":"2025-05-29 12:28:33","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":31473,"visible":true,"origin":"","legend":"\u003cp\u003eHEC-HMS Modelled discharge validation with 2024 March observed flow for Olkeriai Basin\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-6670542/v1/c064dc30ce6b64eae67b4a1f.png"},{"id":83611102,"identity":"75934953-c5c3-451e-af38-ebbb76cbd2ae","added_by":"auto","created_at":"2025-05-29 12:20:34","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":59572,"visible":true,"origin":"","legend":"\u003cp\u003eHEC-HMS Modelled maximum daily rainfall and daily discharges m3/s 1994-2024 and CHIRPS rainfall for steady flow modeling in HEC-RAS.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-6670542/v1/4ab0bcdea452f30b590793aa.png"},{"id":83611096,"identity":"a43e9be2-a258-4a2b-8c19-3781fb92ba74","added_by":"auto","created_at":"2025-05-29 12:20:33","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":34108,"visible":true,"origin":"","legend":"\u003cp\u003eModelled annual Peak (maximum) Discharges for flood frequency analysis.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-6670542/v1/b059c70293f1c31e43d5b95b.png"},{"id":83611098,"identity":"c5a7a9d1-4157-4076-a226-fbd60e0d6126","added_by":"auto","created_at":"2025-05-29 12:20:33","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":21223,"visible":true,"origin":"","legend":"\u003cp\u003eOlkeriai basin Flood Frequency Curve\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-6670542/v1/0fa5e6487b2d61ad4d651851.png"},{"id":83611103,"identity":"e06fe42a-039d-4896-ba7e-52c27de814ab","added_by":"auto","created_at":"2025-05-29 12:20:34","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":708384,"visible":true,"origin":"","legend":"\u003cp\u003eOlkeriai basin 100-year flood depth map.\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-6670542/v1/14e27c6bcc31f4702c22b974.png"},{"id":83611107,"identity":"d2d1a3b2-59de-4748-94dc-0b5b50edd3e7","added_by":"auto","created_at":"2025-05-29 12:20:34","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":314376,"visible":true,"origin":"","legend":"\u003cp\u003eFlood Inundation Index values across Olkeriai River Basin and its catchment area.\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-6670542/v1/c5ea573bab5d716aafa155dc.png"},{"id":83611112,"identity":"b90df0a9-bfa4-4769-943a-f857e58337cd","added_by":"auto","created_at":"2025-05-29 12:20:34","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":259227,"visible":true,"origin":"","legend":"\u003cp\u003eNash-Sutcliffe Efficiency (NSE) Validation graph for the Hybrid model.\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-6670542/v1/84268b27012b26747fae86d6.png"},{"id":83611967,"identity":"2cc21d9b-71c7-4fcf-a41e-b4e7713f3d3f","added_by":"auto","created_at":"2025-05-29 12:28:34","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":374012,"visible":true,"origin":"","legend":"\u003cp\u003eGraph showing the f-score confidence level\u003c/p\u003e","description":"","filename":"10.png","url":"https://assets-eu.researchsquare.com/files/rs-6670542/v1/ce5a684b0d69738bbb31ba6a.png"},{"id":83611965,"identity":"44700068-b4b2-4c65-8914-57e414c3368a","added_by":"auto","created_at":"2025-05-29 12:28:34","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":65723,"visible":true,"origin":"","legend":"\u003cp\u003eScatter plot for the original sample values, biased corrected, regression and an ideal fit line for the optimized model\u003c/p\u003e","description":"","filename":"11.png","url":"https://assets-eu.researchsquare.com/files/rs-6670542/v1/39699ac7bc13f477ff39cb2f.png"},{"id":83611125,"identity":"ae67c5f8-61c0-416a-b571-9ca56f2bda20","added_by":"auto","created_at":"2025-05-29 12:20:34","extension":"png","order_by":12,"title":"Figure 12","display":"","copyAsset":false,"role":"figure","size":38621,"visible":true,"origin":"","legend":"\u003cp\u003eThe residual histogram for mean bias\u003c/p\u003e","description":"","filename":"12.png","url":"https://assets-eu.researchsquare.com/files/rs-6670542/v1/44c18a4eddbe6e673c5f188d.png"},{"id":83611966,"identity":"1052ed03-ecd7-43f4-89cc-52e30d012e4b","added_by":"auto","created_at":"2025-05-29 12:28:34","extension":"png","order_by":13,"title":"Figure 13","display":"","copyAsset":false,"role":"figure","size":260435,"visible":true,"origin":"","legend":"\u003cp\u003eOptimized HEC-RAS flood depth inundation map\u003c/p\u003e","description":"","filename":"13.png","url":"https://assets-eu.researchsquare.com/files/rs-6670542/v1/35bd192458e575cc60b92cad.png"},{"id":83611123,"identity":"e106b21e-5e3f-4ff0-b51f-e6422d896537","added_by":"auto","created_at":"2025-05-29 12:20:34","extension":"png","order_by":14,"title":"Figure 14","display":"","copyAsset":false,"role":"figure","size":279806,"visible":true,"origin":"","legend":"\u003cp\u003eOptimized\u003cstrong\u003e \u003c/strong\u003eflood Inundation map.\u003c/p\u003e","description":"","filename":"14.png","url":"https://assets-eu.researchsquare.com/files/rs-6670542/v1/65fb1817474eacf284ea19b1.png"},{"id":95564011,"identity":"f75c1a2a-9904-4add-8c2c-87e832571ba5","added_by":"auto","created_at":"2025-11-10 16:06:27","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4315943,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6670542/v1/f244b70a-fa3d-46c6-8b69-cf0f3ea0fe30.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Optimization of the HEC-RAS Based Flood Inundation Mapping using Adaptive Neuro- Fuzzy Inference System: A case study of Olkeriai River Basin, Kenya","fulltext":[{"header":"Introduction","content":"\u003cp\u003eFlooding refers to an overflow of water onto the land (Douben, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2006\u003c/span\u003e). The Federal Emergency Management Agency (FEMA) gives a precise definition of a flood as a general and temporary condition of partial or complete inundation of normally dry areas of land or property (Iii, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2005\u003c/span\u003e). Likewise, the Intergovernmental Panel on Climate Change (IPCC) in its Special report on managing the risks of extreme events and disasters to advance climate change adaptation classifies floods as the overflowing of the usual limits of a stream or other watercourse, or the accumulation of water over areas which are not usually covered by water (Ahmed et al, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2007\u003c/span\u003e). All these definitions by the authoritative bodies refer to the main idea of water going beyond its normal limits and spilling over into regions that are normally dry as per (Nordlo, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2016\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAccording to Gaume et al. (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2016\u003c/span\u003e), the two broad categories of flooding are general flooding and flash flooding. While both are characterized by an overflow of water, both are significantly different in the temporal character they display (Now, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). General flooding is generally a longer-term phenomenon that will likely last days or weeks (Munyai et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Conversely, a flash flood is defined by its sudden occurrence, typically brought about by an abnormal or heavy amount of rain in a brief time, typically in under six hours (Feng et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Flash floods are commonly described as turbulent water torrents that can easily devastate riverbeds, urban streets, or mountain canyons and carry everything with them (Khan et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). From the general phenomenon of floods, flood occurrences are usually classified according to their main causes and the surrounding environments in which they take place according to a study by Eskresi, (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Two main categories are fluvial flooding and alluvial flooding (Hamerlynck et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Eskresi, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Mensah, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Riverine flooding, also called fluvial flooding, happens when water in a stream, river, or other watercourse rises and spills over the banks and into the surrounding low-lying areas, which are the natural floodplains (Bernhofen et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). This kind of flooding is most often due to the overflow of heavy or prolonged rain that exceeds the natural capacity of the river channel (Bernhofen et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Fluvial floodings may happen as overbank flooding, with the water level barely exceeding above the stream or river banks, or flash flooding in channels of an already established riverbed from a massive amount of water surging through with little notice (Merz et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2010\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAlluvial flooding, however, is a unique category of flooding occurring on alluvial fans or the same landforms (Piracha, \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Such fan-shaped landforms are developed at the apex, usually where a mountain creek with steep slope meets a relatively flat valley floor. Alluvial flooding involves high-velocity flows, active erosion, sediment transport, and deposition processes, and also irregular flow paths (Piracha, \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Such flooding occurs frequently at the foot of mountain ranges in semi-arid and arid lands (Risi et al., \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Debris flow, which occurs as an intense, concentrated, high-speed water flow carrying a high proportion of sediment and debris, and debris flood with a moderate sediment concentration, may be associated with alluvial fan flooding (Ing, 2010). The processes may move immense amounts of material and are one of the principal sources of hazards (Risi et al., \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2018\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe causes of flooding are varied and function at all geographical scales, ranging from global climatic processes to local environmental and anthropogenic conditions (Mfon et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Understanding these causes is necessary to successfully develop flood inundation and susceptibility maps. Globally, climate change is a leading cause of growing flood risk (Rannie, \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Wahyudi, \u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Global warming produces more intense precipitation events, in that, the warmer atmosphere is able to hold more water, resulting in heavier and more frequent rainfall (Ogega, \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Sea level rise heightens coastal flooding risks (Hallegatte et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Mfon et al. (\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) noted that urbanization of areas at flood risk further aggravates risk by contributing more runoff from impervious surfaces. Looking to regional causative factors, a case study by Mensah (\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) demonstrates coupled causative factors for flood-causing situations. The study notes that heavy rain is one such regular cause and that the El Ni\u0026ntilde;o climatic phenomenon has also caused devastating flooding in East Africa. Poor planning and infrastructure are the primary causes of flood disasters in Ghana and Nigeria (Mensah, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Locally, the direct causes of flooding are excessive rain beyond the capacity of the ground to absorb it, overflows of rivers and lakes, sudden snowmelt, coastal storm surges, failure of dams or levees, and river ice jams (Masese et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Flood risk is also increased locally by urbanization through impervious surfaces and plugged drainage systems (Njogu, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe world is facing an increasing threat from flooding, a natural hazard that has become more frequent and intense in recent decades (Risa, \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Mfon et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Maranzoni et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Floods are the deadliest natural hazards, striking numerous regions in the world each year (Berthelot et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). Between the years 1994 and 2013, floods affected nearly 2.5\u0026nbsp;billion people worldwide, causing more than \u003cspan\u003e$\u003c/span\u003e40\u0026nbsp;billion in damage each year (CRED, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Also, in the past three decades, the average number of flood-related disasters has increased by nearly 35% (Nordlo, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Between 1990\u0026ndash;2022, floods affected more than 3\u0026nbsp;billion people worldwide with 218,353 deaths reported (World, \u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Between 80\u0026ndash;90% of all documented disasters from natural hazards during the past 10 years have resulted from floods, droughts, tropical cyclones, heat waves and severe storms (Piracha, \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). In Africa, the period between 1950 and 2019, small and medium scale floods affected millions of people across the continent affecting nearly seven million people and caused 27,000 deaths (Tramblay et al., \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). The prevalence rates of floods in Kenya stands at 27% and takes 5% of the population affected by disasters as reported by Okaka (\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) and the most commonly affected places by floods are the floodplains of the major rivers such as the lower Tana River, the lower Nzoia River in Budalang\u0026rsquo;i and also the emerging alluvial river systems in arid and semi-arid regions (Njogu, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2021\u003c/span\u003e)\u003c/p\u003e \u003cp\u003eAccording to Morris (\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2014\u003c/span\u003e), the effects of flooding are deep, reaching the natural and human environments in deep and profound ways. The study also notes that the effects can be long-lasting and devastating. From an ecological perspective, flooding devastates habitats in wetlands and forest ecosystems as it leads to soil and riverbank erosion and sedimentation that affect aquatic organisms. Of specific concern is sewage, agricultural runoff, and industrial chemical water pollution, which poses a health risk. Freshwater aquatic organisms may also be affected by flooding. Small floods have beneficial impacts such as recharging groundwater and sedimenting nutrient-dense sediment behind them. Studies by Munyai et al. (\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) and Risa (\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2016\u003c/span\u003e) in Zimbabwe and Nigeria respectively indicated that flooding has detrimental impacts on plant, soil stability, and ecosystem process. The socio-economic damage caused by floods is also catastrophic. CRED (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2015\u003c/span\u003e) analyses the impacts of disasters in that flooding causes enormous infrastructural and property loss and thereby enormous economic damage. The study also notes that entrepreneurial ventures are halted, and agricultural activities are slowed down by destroying crops and animals, people get displaced as an aftermath, waterborne and vector-borne illnesses and psychological effects are the health repercussions.\u003c/p\u003e \u003cp\u003eAccording to Maranzoni et al. (\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), flood inundation mapping relies on a variety of modelling approaches, which can in general be distinguished along a spectrum from physically-based models that reproduce the underlying physical processes to data-driven models that apply statistical and machine learning algorithms to detect patterns and make forecasts. Tshimanga et al. (\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e2016\u003c/span\u003e) also notes that knowledge of the nature of these various types of models is needed to select the most suitable method to an application in flood inundation mapping. Hydrologic Engineering Center-River Analysis System (HEC-RAS), one of the flood simulation tools, is a popular software package for hydraulic computation in river systems (Costabile et al., \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). It applies hydraulic principles to simulate water flow and interaction with natural and human made features. HEC-RAS is applied in floodplain mapping, flood insurance studies, bridge and culvert design, dam break analysis, and channel modification studies (Umamahesh, 2019). It can also be applied for real-time flood forecasting in combination with hydrologic models such as Hydrologic Engineering Center- Hydrologic Modelling System (HEC-HMS). Nordlo (2017) notes that, HEC-RAS has benefits which include the capability to simulate sophisticated hydraulic processes, accommodating a range of flow regimes such as hydraulic structures, ease of use with a Graphic User Interface (GUI) and free availability and that the newer versions of HEC-RAS from version 5.0 to version 6.2 accommodate 2D modeling capabilities. According to Costabile et al. (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2020\u003c/span\u003e), limitations of HEC-RAS include possible numerical instability, requirement of large geometric data, limitation of 1D models to simulate complex flow, requirement of careful calibration and validation, very large computational time for 2D models, and sensitivity to the quality of input data.\u003c/p\u003e \u003cp\u003eHEC-HMS is a computer program hydrologic simulation employed to model precipitation-runoff processes in watershed systems (USACE, \u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e1998\u003c/span\u003e). HEC-HMS models parameters such as precipitation, evapotranspiration, snowmelt, infiltration, runoff, and channel flow as per the HEC-HMS manual USACE, (\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e1998\u003c/span\u003e). The HEC-HMS manual also notes that, this hydrologic model computes flood discharges and produces flow hydrographs that are generally used as an input to hydraulic models such as HEC-RAS. It is employed for flood frequency analysis and also for simulating the effect of land use change on runoff as per the manual USACE, (\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e1998\u003c/span\u003e). The strengths noted from the HEC-HMS manual include large watershed simulation, capability to simulate urban and natural systems, capability to employ many modelling methodologies, integrated GUI, continuous simulation support, and free distribution. Also, the weaknesses noted from the manual are lack of explicit model for hydraulic channel flow complexities, dependency on external GIS packages, complex calibration, and possible limitations in highly steep channels or systems with considerable backwater effects (Guo et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Studies which include Jonkman et al. (\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2008\u003c/span\u003e), Setiawan et al. (\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2018\u003c/span\u003e) have demonstrated the application of HEC-HMS with HEC-RAS and GIS in flood modelling and hazard mapping within defined river bank lines.\u003c/p\u003e \u003cp\u003eAdaptive Neuro-Fuzzy Inference System (ANFIS) is a hybrid artificial intelligence that combines neural networks and fuzzy logic to model complex, nonlinear relationships from data (Walia, \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). ANFIS is data-trained to tailor the parameters of a fuzzy system, generally a Takagi-Sugeno model (Walia, \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). ANFIS has been extensively used in flood susceptibility mapping to estimate flood probability in terms of topography, meteorology, and land use (Ullah, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). ANFIS is also applied to river flow rate and water level flood forecasting (Al-hmouz et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). Walia (\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2015\u003c/span\u003e) notes the benefits and limitations associated with ANFIS, the benefits noted include the ability to learn nonlinear relationships, integration of learning and reasoning, flexibility, dealing with uncertainty, and relatively fast learning rates. Limitations noted are based on training data quality dependency, curse of dimensionality as a result of a high number of input variables, optimal parameters being difficult to select, absence of apparent physical sense, danger of getting stuck in local minima, and variation of performance depending on architecture and training algorithm (Bui et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Studies like Al-hmouz et al. (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2012\u003c/span\u003e) and Bui et al. (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2018\u003c/span\u003e) have established the efficiency of ANFIS in flood susceptibility mapping and forecasting in different regions. Additionally, geospatial technology is useful in flood control and mapping flood risk of hazard areas, i.e., it can be applied prior to floods in areas prone to floods in order to prepare cases of future events, during floods to track floods and post-floods for damage evaluation and countermeasure measures (Saudi, \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn the present study, GIS integrated with hydraulic model, HEC-RAS are used together with remote sensing to simulate results for flood inundation extent. The main objective was to model flooding extent of Olkeriai River basin for the steady flow analysis by optimizing the outputs of HEC-RAS model using Adaptive Neuro-Fuzzy Inference System (ANFIS) machine learning algorithm. The objective was to improve the spatial precision and predictive reliability of flood mapping in this extreme condition so that land-use planning and disaster preparedness are formulated according to the worst-case but feasible scenario.\u003c/p\u003e \u003cp\u003eThe need to model for steady flow conditions is the reason why HEC-RAS was chosen for this study as this model is particularly strong is such conditions therefore allow more accurate representation of flood events as noted by Costabile et al. (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) which also shows that HEC-RAS has the ability to generate flood plain maps and flood extent maps with high spatial accuracy. However, HEC-RAS uses limited data inputs which includes river discharge data, land use land cover data, digital elevation model of the area and the soil characteristics of the study area which limits the model\u0026rsquo;s capacity to account for broader spatial variability and interactions across the floodplain (Al, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). It is also noted by the same study that HEC-RAS has a limitation of confined simulation within the defined bank lines, its spatial limitation to the river corridor presents a knowledge gap in landscape-scale flood hazard mapping. These shortcomings limit the comprehensive mapping of flood zones in the area. By incorporating multiple floods contributing factors which influence flooding beyond the river corridor as input variables for the model, ANFIS machine learning algorithm was therefore incorporated for this study to optimize the outputs obtained from HEC-RAS and to accurately predict and to extend flood inundation mapping coverage beyond the riverbanks, to map flood extents across the catchment area. This improves the accuracy of flood inundation mapping by taking advantage of ANFIS capability in detecting intricate relationships among model inputs and flood extents observed. This method brings a paradigm change in flood hazard mapping that further helps in formulating flood mitigation strategies. Flood hazard awareness and avoidance will improve the resilience of smallholder farming families by safeguarding their crops and livelihood from damages caused by floods.\u003c/p\u003e"},{"header":"Methodology","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\n \u003ch2\u003eDescription of the study area\u003c/h2\u003e\n \u003cp\u003eThe Olkeriai basin is located in Kajiado county in Kenya and lies between longitude 37⁰25\u0026rsquo;00\u0026rdquo;E to 30⁰ 50\u0026rsquo;00\u0026rdquo;E and latitude 1⁰45\u0026rsquo;00\u0026rdquo;S to 2⁰30\u0026rsquo;00\u0026rdquo;S with an approximate area of 1524 Km\u003csup\u003e2\u003c/sup\u003e (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e). The river is seasonal and sandy with a length of approximately 100 km and forms one of the tributaries that drain its water into river Athi basin. The basin, is semi-arid with bimodal rainfall, with short rains occurring between October and December and long rains from March to May, ranging from 300 to 800 mm annually. The temperature ranges from high temperatures of 28\u0026deg;C to 35\u0026deg;C and low temperatures of 17\u0026deg;C to 22\u0026deg;C, with January and February being the hottest months (Uker et al., \u003cspan class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e\n \u003cp\u003e60 km stretch of the river where agriculture is mainly practiced and experiences riparian flooding near Mashuuru town was used for analysis and modelling. The basin also features land with semi-arid savannah plains intermixed with seasonal river valley areas (Matsaba et al., \u003cspan class=\"CitationRef\"\u003e2021\u003c/span\u003e). Geology in this region includes deposits from volcanic processes of Quaternary and Pleistocene together with basement rocks, which in turn produce types of soil as black cotton soil and sandy loams (Uker et al., \u003cspan class=\"CitationRef\"\u003e2022\u003c/span\u003e). This region is socio-economically dependent on pastoralism and plough agriculture dependent on the semi-arid climatic conditions coupled with unpredictable regimes of rainfall (Duker et al., \u003cspan class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e\n\u003c/div\u003e\n\u003ch3\u003eDatasets used for flood inundation mapping using HEC-HMS, HEC-RAS and ANFIS\u003c/h3\u003e\n\u003cp\u003eThe study utilised several datasets as presented in Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e, which are, European Space Agency (ESA) Sentinel-2 (10-meter 2024 imagery) for curve number/Manning\u0026apos;s coefficients and Land Use/Land Cover (LULC) classification, Global Hydrologic Soil Groups (A-D classification) from Distributed Active Archive Center - Global Hydrologic Soil Groups (DAAC-GHSOGS) for Modelling in HEC-HMS for estimation of runoff, Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) daily data as input in HEC-HMS meteorological component, European Space Agency (ESA) 30- metre spatial resolution Copernicus Digital Elevation Model (DEM) for mapping terrain and flow in the HEC-HMS and HEC-RAS models, and Water Resources Authority (WRA) discharge data for flood modelling in the HEC-HMS model and calibration. Elevation, slope, lineament density and Topographical Wetness Index (TWI) were extracted from a 30m spatial resolution Copernicus DEM and used as inputs for the ANFIS model development. The LULC, Normalized Difference Vegetation Index (NDVI) data maps and Manning\u0026rsquo;s Coefficient which are also the input datasets for the ANFIS model development were created by carrying out supervised classification of a 10m spatial resolution Sentinel-2image acquired using Google Earth Engine platform from Copernicus.\u003c/p\u003e\n\u003cp\u003e\u003c/p\u003e\u0026nbsp;\u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eDatasets used, their description and their relevance in the study.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eDataset\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSource\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSpecification\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eRelevance\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSentinel-2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEuropean Space Agency (ESA)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMultispectral 2024 imagery,\u003c/p\u003e\n \u003cp\u003e10m resolution\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLULC classification to derive CN \u0026amp; Manning\u0026rsquo;s roughness coefficients. Also used to derived NDVI which is an input data for ANFIS model development.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSoils(GHSOGS)\u003c/p\u003e\n \u003cp\u003eGlobal Hydrologic Soil Groups\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDistributed Active Archive Center (DAAC)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHydrologic soil group classification (A-D),\u003c/p\u003e\n \u003cp\u003eglobal coverage,\u003c/p\u003e\n \u003cp\u003ebased on soil texture\u0026amp; infiltration rates\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEssential for creating the curve number (CN) map in HEC-HMS to estimate runoff potential based on soil moisture retention in the basin.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRainfall\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCHIRPS (Climate Hazards Group InfraRed Precipitation with Station data)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDaily rainfall data\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eProvides meteorological input for HEC-HMS to simulate runoff and discharge.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDEM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCopernicus Programme (European Space Agency)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e30m resolution digital elevation model, derived from satellite altimetry\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eUsed for terrain mapping, watershed delineation, flow direction, and stream network generation in HEC-HMS and HEC-RAS to model hydrologic and hydraulic responses. Also used to extract elevation, slope, lineament density and TWI which are the input data for ANFIS model development.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDischarge\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWater Resources Authority (WRA)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eObserved discharge\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCalibrates Observed discharge which serves as input in HEC-RAS model. Also, an input in ANFIS model development.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003c/p\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003ch3\u003eWorkflow for flood inundation mapping using HEC-HMS, HEC-RAS and ANFIS\u003c/h3\u003e\n\u003cp\u003eThe flowchart in Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e shows the methodology for flood inundation mapping in Olkeriai River Basin with hydrological and hydraulic modelling using HEC-HMS and HEC-RAS respectively, ANFIS model development and the process of optimization of HEC-RAS outputs. The process was carried out first, through terrain delineation and watershed setup to create subbasins and reaches in the basin. Supervised pixel based random trees classification was then carried out to produce LULC map which was used together with soils data to create curve number map. These were then utilized as inputs to HEC-HMS for basin model configuration with addition of lag time. The HEC-HMS model was then calibrated with the observed Olkeriai River basin flow data for the March 2024 flooding event against daily discharge at Mashuuru to generate daily discharge and annual maximum data for Olkeriai basin. With annual maximum discharges obtained by the HEC-HMS model, Flood Frequency Analysis (FFA) was then done in the basin to determine peak discharges for the 100-year flood return period for the basin. Steady flow analysis was then simulated using HEC-RAS to obtain flood velocity and depth maps of the basin flood plains using the steady discharge from the 100-year return period. The outputs from HEC-RAS together with the modelled daily discharge data, elevation data, TWI data, slope, lineament density, soil data, LULC data for the year 2024 and the NDVI data and the HEC-RAS output to train and validate ANFIS model. Optimization process was achieved by overlaying the HEC-RAS outputs on the flood inundation index map based on ANFIS machine learning algorithm followed by extraction of corresponding pixel values in HEC-RAS and ANFIS outputs then an optimized flood depth inundation map produced.\u003c/p\u003e\n\u003cp\u003eThe flow discharge data for the March 2024 floods for Olkeriai River basin was used to calibrate and validate the HEC-HMS for this study. Steady flow analysis was utilized in this study because of its applicability in simulation under peak flood conditions where the flow conditions are considered to be constant with respect to time (Shabir, \u003cspan class=\"CitationRef\"\u003e2016\u003c/span\u003e). Steady flow analysis is a simple but effective method for flood depth and area determination in worst flood conditions (Access, \u003cspan class=\"CitationRef\"\u003e2019\u003c/span\u003e). 100-year return period flood was employed since it is a severe flood with 1% probability of occurrence in any year, according to Conventional Engineering and Risk Management Practice for the Estimation of Flood Hazards. The 100-year flood extent results were optimized using ANFIS machine learning algorithm, as it is the situation that poses maximum threat to life, property, and agriculture in the Olkeriai River Basin.\u003c/p\u003e\n\u003ch3\u003eModelling Olkeriai River discharge data using HEC-HMS\u003c/h3\u003e\n\u003cp\u003eHEC-HMS was first used to simulate Olkeriai river basin flow discharge data for the steady flow analysis with HEC-RAS. The model was coupled with a 30 years period 1994\u0026ndash;2024 daily rainfall dataset from CHIRPS with a Curve Number (CN) map derived from global hydrological soil groups (DAAC) and land use land cover map, lag time estimates, and topographical data obtained from Copernicus 30m DEM.\u003c/p\u003e\n\u003cp\u003eThe Olkeriai River basin\u0026rsquo;s model was delineated by marking the watershed and defining its physical characteristics with the help of Copernicus 30m DEM. The DEM provided high-resolution data regarding flow direction, accumulation, and stream networks for the accurate representation of the hydrologic response of the basin. A total of 15 subbasins were generated and one was merged to as they were too small in area. The CN map developed using global hydrological soil groups and LULC map, created according to the Soil Conservation Service (SCS) method to estimate runoff potential, was averaged for each sub basin through zonal statistics. Lags, defined as the time difference between peak rainfall and peak discharge, were estimated by empirical relationships using the watershed lag method by United States Department of Agriculture (USDA), Natural Resources Conservation Service (NRCS) (USDA and NRCS 2008), as a function of basin characteristics using Eqs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e and 2 as in Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e\n\u003cdiv id=\"Equ1\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equ1\" name=\"EquationSource\"\u003e$$\\:L=0.6{T}_{c}$$\u003c/div\u003e\n \u003cdiv class=\"EquationNumber\"\u003e1\u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv class=\"gridtable\"\u003e\n \u003cdiv align=\"left\" class=\"colspec\"\u003e\u003cbr\u003e\u003c/div\u003e\u0026nbsp;\u003ctable id=\"Taba\" border=\"1\"\u003e\n \u003ccolgroup cols=\"3\"\u003e\u003c/colgroup\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{T}_{c}=\\:\\frac{{\\mathcal{l}}^{0.8}{(S+1)}^{0.7}}{1140{Y}^{0.5}}\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" style=\"width: 5.7118%;\"\u003e\u0026nbsp;\u003cbr\u003e\n \u003cp\u003e(2)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Equ2\" class=\"Equation\"\u003e\n \u003cp\u003ewhere: L is lag, h; \\(\\:{T}_{c}\\) is the time of concentration, t; \u0026lambda;\u0026thinsp;=\u0026thinsp;flow length, ft; Y is the average watershed land slope, percent; S is the maximum potential retention obtained using Eq. 3\u003c/p\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equ2\" name=\"EquationSource\"\u003e$$\\:S=\\:\\frac{1000}{\\text{c}\\text{n}{\\prime\\:}}-10$$\u003c/div\u003e\n \u003cdiv class=\"EquationNumber\"\u003e3\u003c/div\u003e\n\u003c/div\u003e\n\u003cp\u003ewhere: cn\u0026prime; is the retardance factor which is approximately the same as the curve number (CN).\u003c/p\u003e\n\u003cp\u003e\u003c/p\u003e\u0026nbsp;\u003ctable id=\"Tab2\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eCurve Number, Slope, Time of Concentration and Lag of each subbasin in Olkeriai River Basin\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSubbasin\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCN\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSlope\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eT\u003csub\u003ec\u003c/sub\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eLag (min)\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSubbasin-1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e87.76\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.39\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2.62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e94.21\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSubbasin-3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e89.67\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e34.09\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSubbasin-4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e87.46\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3.20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e115.14\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSubbasin-5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e86.89\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.51\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2.96\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e106.38\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSubbasin-6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e87.36\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.45\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e36.07\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSubbasin-7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e87.58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.42\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.96\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e70.74\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSubbasin-8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e86.66\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.54\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e112.40\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSubbasin-9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e87.36\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.45\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.40\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e50.48\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSubbasin-10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e88.52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.77\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e63.76\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSubbasin-11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e88.06\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.36\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2.27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e81.64\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSubbasin-12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e87.50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e58.16\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSubbasin-13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e88.75\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e48.00\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSubbasin-14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e91.62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.91\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.97\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e34.79\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSubbasin-15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e87.38\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.44\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3.61\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e129.85\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003c/p\u003e\n\u003cp\u003eThe Soil Conservation Service Curve Number (SCS-CN) was chosen as the loss method in the hydrologic modelling because it could accommodate the soil moisture and land use characteristics. This method of estimating runoff involves rainfall, soil type, and land cover with less input parameters required curve number and initial abstraction. The SCS-CN procedure in HEC-HMS estimates the runoff based on assigning CN values to LULC and HSGs with Olkeriai basin having soil groups C, C/D, D, and D/D. CN values were assigned based on the HEC-HMS manual USACE (\u003cspan class=\"CitationRef\"\u003e1998\u003c/span\u003e), these CN values varied between 30 and 100, with higher numbers representing more runoff as shown in Table \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e. After preparation of the curve number map using the LULC and the hydrological soil groups, the average curve number for each subbasin was calculated using the zonal statistics tool in ArcGIS pro software.\u003c/p\u003e\n\u003cp\u003e\u003c/p\u003e\u0026nbsp;\u003ctable id=\"Tab3\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eTable for Assigning SCS Curve Numbers for HEC-HMS\u003c/p\u003e\n \u003cdiv class=\"Credit\"\u003e\n \u003cp\u003e(Source: USACE (\u003cspan class=\"CitationRef\"\u003e1998\u003c/span\u003e))\u003c/p\u003e\n \u003c/div\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eLULC Type\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eHSG C\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eHSG C/D\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eHSG D\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAgriculture\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e79\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e85.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e91\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSand\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e74\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e75.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e77\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDeveloped\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e88\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e91.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e94\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBare land\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e84\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e87.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e91\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eThick Forest\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e77\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e81.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e88\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSparse Forest\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e88.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e91\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eShrubland\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e81\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e86.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e92\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWetlands\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e92.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e100\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003c/p\u003e\n\u003cp\u003eThe transform method used was the Snyder Unit Hydrograph method. This method uses parameters such as basin lag time and peak rate factor, (USACE, \u003cspan class=\"CitationRef\"\u003e1998\u003c/span\u003e), which were calibrated using observed flow data. For routing, the Muskingum method as described by Cunderlik, (\u003cspan class=\"CitationRef\"\u003e2004\u003c/span\u003e) was applied to simulate water movement through river channels, balancing accuracy and computational efficiency.\u003c/p\u003e\n\u003cp\u003eA meteorological component was developed from the 30-year time series rainfall data obtained from CHIRPS and treated to account for the spatial and temporal distribution within the study area. Hydrologic simulation in HEC-HMS was performed considering steady flow conditions. Seventy percent of the observed discharge data for the March 2024 flooding event in the Olkeriai River basin was used to calibrate the model, also, manual calibration was done by adjustment of parameters such as SCS unit hydrograph values, lag times, graph type, and Muskingum x and k coefficients (Costabile et al., \u003cspan class=\"CitationRef\"\u003e2020\u003c/span\u003e). Univariate Gradient optimization simulation was created to further improve the accuracy of the model through automated calibration (Cunderlik, \u003cspan class=\"CitationRef\"\u003e2004\u003c/span\u003e). The remaining 30% of the discharge data was used to validate the model.\u003c/p\u003e\n\u003ch3\u003eFlood frequency analysis (FFA) for steady flow simulations\u003c/h3\u003e\n\u003cp\u003eThe process of steady flow analysis was done by first designing steady peak discharge related to floods that have 100-year probability of occurrences. This is useful in modelling the worst-case scenario floods that are extremely severe in case of occurrence. With annual maximum discharges obtained by the HEC-HMS model, Flood Frequency Analysis (FFA) was done in the basin to determine peak discharge of this flood return period.\u003c/p\u003e\n\u003cp\u003eThe log-persons type III probability distribution was used in calculation of discharges corresponding to a given probability of exceedance and it has been clearly defined and implied by Mendez et al. (\u003cspan class=\"CitationRef\"\u003e2023\u003c/span\u003e). The probability of exceedance is defined as the inverse of the return period as per Head et al. (\u003cspan class=\"CitationRef\"\u003e2023\u003c/span\u003e) for example the 100-year return period corresponds to a 1% probability of exceedance which translates to 1% chance of happening at any year, calculated as in Eq.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e.\u003c/p\u003e\n\u003cdiv id=\"Equ3\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equ3\" name=\"EquationSource\"\u003e$$\\:P=\\:\\frac{1}{T}$$\u003c/div\u003e\n \u003cdiv class=\"EquationNumber\"\u003e4\u003c/div\u003e\n\u003c/div\u003e\n\u003cp\u003eWhere P is the probability and T is the return period.\u003c/p\u003e\n\u003cp\u003eThis method has been applied successful in flood frequency analysis where it has been used in Kenya by Lang\u0026rsquo;at et al. (2019) in Tana River basin and Houessou-dossou et al. (\u003cspan class=\"CitationRef\"\u003e2023\u003c/span\u003e) for FFA analysis in Narok town.\u003c/p\u003e\n\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\n \u003ch2\u003eSteady flow simulation using HEC-RAS\u003c/h2\u003e\n \u003cp\u003eIn steady flow simulation using HEC-RAS, the 30m Copernicus DEM was used to generate terrain data. Then the river center line, bank lines, flow paths and cross sections was manually generated. Then manning roughness coefficients were assigned to the river bed and bank lines based on land use land cover. A steady data profile was created where discharge from FFA were used for the 100-year return period profile. Steady flow analysis was simulated to obtain flood extent and depth maps of the basin flood plains. For result validation, the produced flood extent maps were compared to field observations and historical flood extent information gathered by interviewing local people during a field survey carried out in the region.\u003c/p\u003e\n\u003c/div\u003e\n\u003ch3\u003eDevelopment of Adaptive Neuro-Fuzzy Inference System (ANFIS) model\u003c/h3\u003e\n\u003cp\u003eAdaptive Neuro-Fuzzy Inference System (ANFIS) was implemented to model flood inundation index map within the Olkeriai River Basin based on HEC-RAS model outputs and geospatial flood conditioning factors. The model incorporated the result of the 100-year flood simulation by HEC-RAS and the flood causative factors to generate a spatially continuous flood depth inundation map, which was utilized in the optimization process.\u003c/p\u003e\n\u003cp\u003eFlood depth output from HEC-RAS steady-flow simulation for return period of 100 years was used as the dependent variable in ANFIS model. HEC-RAS flood depth output was exported as a raster layer and resampled to spatial resolution of 30 meters, matching Landsat-derived NDVI and DEM resolution. The Universal Transverse Mercator (UTM), Zone 37S, World Geodetic System 1984 (WGS 84) datum was used as the base coordinate reference system for all the spatial layers in order to have consistency and spatial integrity. Six independent variables (flooding conditioning factors) were used because of their established relevance in flood inundation studies in the literature, they include the following; elevation, slope, mean annual rainfall, stream flow, NDVI, TWI.\u003c/p\u003e\n\u003cp\u003eAll raster layers were reprojected, clipped to the area of study, resampled at 30 m resolution, and co-registered through the use of the SNAP toolbox in ArcGIS pro software. To standardize the inputs to be fed into the ANFIS model, min-max normalization was applied on every continuous input, converting the values into normalized range of 0 to 1 using the formulae shown in Eq.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e. This normalization method provided evenly weighted scaling to all the inputs so that the neural-fuzzy model is able to converge efficiently during training. This method has already been applied and shown successful in flood modelling by (Palakondi, 2017).\u003c/p\u003e\n\u003cdiv id=\"Equ4\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equ4\" name=\"EquationSource\"\u003e$$\\:{y}_{n}=\\frac{y-{y}_{1}}{{y}_{2}-y}$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e5\u003c/div\u003e\u003c/div\u003e\u003cp\u003eWhere; \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{y}_{n}\\)\u003c/span\u003e\u003c/span\u003e is normalized value (transformed within 0and 1 range), \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:y\\)\u003c/span\u003e\u003c/span\u003e the original influencing factor value at the sampling point, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{y}_{1}\\)\u003c/span\u003e\u003c/span\u003e is the minimum value of a factor across all locations and, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{y}_{2}\\)\u003c/span\u003e\u003c/span\u003e represents the maximum value of a factor across locations respectively.\u003c/p\u003e\u003cp\u003eFor training the ANFIS model, a sample dataset was created by extracting values from each of the input raster layers and the corresponding flood depth raster. 10,000 spatial sample points were generated within the study area using a stratified random sampling process within ArcGIS, with such, a representative coverage was obtained across different landform and flood-prone areas. For every sample point, the input pixel values of the six input variables were extracted, as well as the corresponding flood depth value as the output target. The created dataset (10,000 records \u0026times; 7 variables) was exported in Comma-Separated values (CSV) format and imported as an input to the ANFIS model in Python. The data set was divided into 70% for training (7,000 records) and 30% for testing (3,000 records). The training data set was utilized to create the neuro-fuzzy inference rules, whereas the test data set was reserved for model evaluation and validation.\u003c/p\u003e\u003cp\u003eANFIS model development was executed using Python code of local code in the available open-source platforms created using NumPy and TensorFlow. A Gaussian membership function was applied to all the input variables to be utilized in the fuzzy set representation, and the system employed a first-order Sugeno-type fuzzy inference system (Walia, \u003cspan class=\"CitationRef\"\u003e2015\u003c/span\u003e). The number of membership functions in every input was manually tuned to obtain the best complexity and accuracy level. Hybrid optimization algorithm with least-squares estimation for linear parameters and backpropagation in order to optimize nonlinear parameters was employed until convergence and stability of the validation error was attained using the procedure of Walia (\u003cspan class=\"CitationRef\"\u003e2015\u003c/span\u003e). Following successful training, the ANFIS model was executed using the whole raster stack (6 layers) over the study area. The trained model was input with all pixels\u0026apos; values in order to calculate a flood inundation index score, generating a continuous raster surface between 0 (low inundation) and 1 (high inundation).\u003c/p\u003e\u003cp\u003eThe performance of the model was evaluated using the test dataset where calculated flood inundation scores were compared against the observed depth of flood. Model performance metrics which were used are; Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Coefficient of Determination (R\u0026sup2;). Furthermore, the flood inundation map was cross-validated with the Receiver Operating Characteristic (ROC) curve, and observed flood areas (from the 100-year simulation) were used as reference. The Area Under the Curve (AUC) was estimated, where AUC\u0026thinsp;\u0026gt;\u0026thinsp;0.80 was taken as an indication of high spatial prediction performance following a criterion by Janssens, (\u003cspan class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e\u003ch3\u003eOptimizing HEC-RAS Simulation Using ANFIS Data to develop a hybrid model\u003c/h3\u003e\u003cp\u003eThe sample datasets were obtained by overlaying HEC-RAS flood inundation map on the developed flood inundation index map based on ANFIS then extracting corresponding pixel values that are matching using the spatial analyst extension tool in ArcGIS pro software. HEC-RAS data provided outputs that are restricted along the river geometry which include variables between flood depth and velocity maps. The flood inundation index map based on ANFIS provided a broader output coverage since it extends to cover the whole catchment area through its ability to integrate multiple environmental and hydrological factors in its prediction. The specific flood depth values from HEC-RAS were matched to correspond with values from ANFIS flood inundation index map.\u003c/p\u003e\u003cp\u003eIn dealing with erroneous or missing data values, mean substitution for numerical data and interpolation techniques for other data types was used. The Z-score statistical method with 95% confidence level was used to detect outliers for removal because these points had the potential to distort the model\u0026apos;s performance results. Data cleaning was done to minimize data distortion caused by missing or incorrect entries. Stratified random sampling method in ArcGIS was used for dataset selection to pick data points evenly across all conditions in the river catchment from the HEC-RAS and ANFIS datasets. Three hundred data points were effectively collected across all depths of floodwaters along with various types of terrain and hydrological contexts essential for building an effective model. The datasets were divided into two distinct sets, one of which trained the model and a second sample set to test model performance. Training dataset consisted of 70% of three hundred sample points, this provided the model with the ability to know how to make model predictions based on HEC-RAS flood simulation. Test part of the model comprised 30% of data to test its performance in coping with unknown data. Stratified sampling was employed to ensure data distribution in both subsets.\u003c/p\u003e\u003cp\u003eGeographically Weighted Regression (GWR) was utilized to correlate HEC-RAS flood depth results to spatial ANFIS model outputs. The model was trained on 70% of the data and was implemented by choosing an optimal kernel and bandwidth by using cross-validation. Bias correction was done to overcome the HEC-RAS model restriction that operated only within defined river geometry. The evaluation for identifying systematic biases in data proceeded through the utilization of Bland-Altman Analysis as per Fernandez (\u003cspan class=\"CitationRef\"\u003e2009\u003c/span\u003e), that measured differences in the HEC-RAS values ANFIS developed map. The deviations of the mean values from zero indicates that HEC-RAS outputs had regular deviations from ANFIS values either in positive or negative direction. Quantile Mapping and Bias Adjustment Factors were applied in analysis for bias correction process. Quantile Mapping as stated Hamill (\u003cspan class=\"CitationRef\"\u003e2018\u003c/span\u003e) was employed as a technique of matching the cumulative distribution functions (CDF) of outputs from HEC-RAS and ANFIS data. The corrected HEC-RAS outputs were tuned with this method by aligning them with the statistical requirements of the ANFIS dataset. Bias Adjustment Factor was calculated by checking systematic differences among the original datasets. Bias Adjustment Factor allowed the removal of residual biases from HEC-RAS output data through its application.\u003c/p\u003e\u003cp\u003eFor validation of the model, 30% of the randomly picked data points was used to validate calibrated and bias-corrected model. The metrics which was used to evaluate the performance of the optimized model included the Coefficient of Determination (R\u003csup\u003e2\u003c/sup\u003e), Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Nash- Sutcliffe Efficiency (NSE) and the mean bias error. To improve model performance, the analysis process was repeated several times while fine tuning the hyperparameters and optimization procedures.\u003c/p\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003eFinal Optimized HEC-RAS based Flood Inundation Simulation\u003c/h2\u003e\u003cp\u003eThe process involved utilizing the trained and optimized model for HEC-RAS output simulation across the entire river catchment region. The model used ANFIS output data to predict water depth extent and hydraulic conditions throughout the whole catchment. The simulated output from HEC-RAS optimization covered the entire area of the river catchment solving the limitation of HEC-RAS of being confined within the defined river geometry.\u003c/p\u003e\u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eSimulated river flow discharge using HEC-HMS Model\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOn running the HEC-HMS hydrological simulation, a Root Mean Square Error (RMSE) standard deviation of 0.411 and a Nash-Sutcliffe Efficiency (NSE) of 0.848 was obtained after final calibration of the model and can be illustrated in Fig. 3. This enabled the simulation of the 30 years rainfall to obtain credible discharges that are crucial in modelling flood inundation extents for the study.\u003c/p\u003e\n\u003cp\u003eThe HEC-HMS model successfully modelled daily discharges from the year 1994 to 2024 as shown in Fig. 4\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003ein order to identify extreme hazard flooding events in the Olkeriai basin and for steady flow modelling in HEC-RAS. The model results showed close corelations between extreme rainfall and high discharges that lead to flooding of the river as depicted in Fig. 4. \u0026nbsp;Maximum annual discharge data was also modelled as shown in Fig. 5 for use in flood frequency analysis.\u003c/p\u003e\n\u003cp\u003eFlood Frequency Analysis was then performed using the annual discharge data shown in Fig. 5. This was to determine peak discharge for 100- year flood return period for the basin which was to be used for steady flow analysis in HEC-RAS. The determined peak discharge is illustrated by the curve in Fig. 6. The peak discharge obtained for the 100-year return period was 3749.3 m\u003csup\u003e3\u003c/sup\u003e/s. The discharge data was now ready for modelling of flood extents in the HEC-RAS model.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSimulated flood inundations for the 100-year flood using HEC-RAS Model\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSteady flow analysis was simulated to obtain flood inundation extent and depth maps of the basin flood plains using the steady data profile which was created where discharge from FFA were used for the 100-year; 3749.3 m3/s return period profile. The 100-year flood extent map shown in Fig. 7 flood covers 37.27 km\u0026sup2; with an average depth of 3.22 meters. From the river corridor, small water extent coverage is seen on the Northern region of the study area which is a high elevated region while large extents are observed on the lower region of the study area. Higher water depth values are also observed on the southern region of the study area and along the river channel indicated by deep green colour coding. This agrees with the observation for the water inundation coverage that a larger extent is observed on the Southern region of the map, this implies that the southern region of the map is more susceptible to flooding due to its low elevation. Low depth values are observed mainly in the northern region of the study area.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFlood inundations index mapping using Adaptive Neuro-Fuzzy Inference System (ANFIS)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo evaluate the performance of the trained ANFIS model, observed flood depths (from 100-year simulation with HEC-RAS) were compared to ANFIS-predicted values. The following statistical measures below were calculated: Root Mean Square Error (RMSE) of 0.092, Mean Absolute Error (MAE) of 0.090, and Coefficient of Determination (R\u0026sup2;) of 0.960 represented in Table 4. Low RMSE values indicates that the model performed well with low prediction errors. These results represent good model performance, with small error values and a high R\u0026sup2; score representing the model capability to explain flood depth variability accurately.\u003c/p\u003e\n\u003cp\u003eThe performance of the model in classification accuracy was also tested through a Receiver Operating Characteristic (ROC) curve. This was achieved through the development of predicted flood inundation index scores against flood inundation depth values from the 100-year flood inundation map. As shown in Table 4, the derived Area Under the Curve (AUC) was 0.910. AUC value more than 0.90 indicates that the model strongly discriminates flood-prone and non-flood-prone regions over the study area.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 4\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eANFIS model performance metrics and their corresponding values\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003e\u003cspan id=\"_Toc192696090\"\u003eModel performance metrics\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003eMetric value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003e\u0026nbsp;Root Mean Square Error\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003e0.092\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003eMean Absolute Error\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003e0.090\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003eCoefficient of Determination\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003e0.961\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003eArea Under Curve\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003e0.910\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eThe trained ANFIS model was then used spatially to produce a continuous flood inundation index map of the Olkeriai River Basin. The flood causative factor values of each pixel were used as inputs to the trained model to calculate the flood inundation indices between 0 (very low) and 1 (very high). The resulting flood inundation index map is presented in Fig 8. Higher depth values are seen along the river geometry on the central region of the study area.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eOptimized HEC-RAS based Flood Inundation using ANFIS data\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe hybrid model of ANFIS and HEC-RAS achieved an NSE value of 0.944 together with Coefficient of Determination, R\u003csup\u003e2\u003c/sup\u003e value of 0.944. This indicates 94.4% of the observed data variance thereby demonstrating excellent reliability and prediction abilities as shown in Fig. 9.\u003c/p\u003e\n\u003cp\u003eFig. 10 is graphical representation showing model predictions of water depth between the HEC-RAS model along with the ANFIS-calibrated version against validation points while displaying varying degrees of uncertainty intervals at 68% and 95% and 99.7%. The calibrated ANFIS model appears as the blue line alongside the original HEC-RAS model that shows its outputs as the orange line. The alignment between the two curves reveals strong agreement especially in the shaded uncertainty areas.\u003c/p\u003e\n\u003cp\u003eBased on results in Fig. 11, Fig. 12 and Table 5\u003cstrong\u003e,\u0026nbsp;\u003c/strong\u003ethe hybrid model shows high reliability in its validation stages. The developed predictive model produces Coefficient of Determination value of 0.944, Root Mean Square Error (RMSE) values of 0.445 and Mean Absolute Error (MAE) of 0.337 which illustrate its predictive power through low error estimation. Validation process used a total of 275 observation points from a sample size of 300 observed data points, 25 data points were removed since they were outliers and could bring skewedness in the prediction. Results from the Mean Bias Error evaluation show 0.098 as the calculated value indicating no major distortion that would impact the model performance. The overall model validation results prove that the model is well-optimized, accurate, and highly suitable for prediction.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 5\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eSummary of model validation metrics and their corresponding values for the optimized model\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003eModel validation metric\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003eMetric value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003eRoot Mean Square Error\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003e0.445\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003eMean Absolute Error\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003e0.337\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003eCoefficient of Determination\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003e0.944\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003eNash-Sutcliffe Efficiency\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003e0.944\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003eMean Bias Error\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 301px;\"\u003e\n \u003cp\u003e0.098\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eFig. 13, optimized HEC-RAS flood depth extent map shows predicted depth values in the entire river catchment by spatial distribution ranging from 0.09m to 10.09m. Low depth values showed regions which have low inundations and lower risk of flooding and mostly appear on the upper region of the basin that has higher elevation. Higher depth values on the other hand that indicate regions that are at greater risk of flooding within the basin occupying most of the region along the river Olkeriai corridor and on the regions with low elevation.\u003c/p\u003e\n\u003cp\u003eThe predicted values were reclassified into three categories based on Jenks method of flood hazard risk categorization as shown by Indexes (2021) and then visualized in GIS as shown in Fig. 14\u003cstrong\u003e.\u0026nbsp;\u003c/strong\u003eThe gradient color scheme presents prediction intensity variations through its green-to-red color transition. Estimated values that are red or orange appear mostly in river corridors and lower elevation zones thus marking potential vulnerable flood areas and high-flow zones. The green sections denote regions with reduced predicted values along with the upland areas and areas less prone to flooding.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 6\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eshows the area extent of the reclassified map based on the inundation indices. The highest probability of flooding exists in moderate-risk zones occupying 41.09% of the catchment area mostly located in low elevated areas in the northern region of the study area. These areas need strict flood mitigation through barriers and improved drainage systems along with built structure controls. High-risk zones which account for 39.23% of the total area which occupies the area along the river corridor and low elevated zones in the southern part of the area shows occasional yet seasonal flood patterns during extreme weather conditions. The area of the basin amounting to 19.68% occupied low-risk region as these locations either rest at elevated heights or possess good drainage characteristics mostly seen in the western and North western region of the area.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 6\u003c/strong\u003e Statistics showing area extent in percentage of reclassified optimized flood risk map.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 200px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eFlood risk category\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 200px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePixel count\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 200px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePercentage cover\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 200px;\"\u003e\n \u003cp\u003eHigh risk zone\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 200px;\"\u003e\n \u003cp\u003e669331\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 200px;\"\u003e\n \u003cp\u003e39.23\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 200px;\"\u003e\n \u003cp\u003eModerate risk zone\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 200px;\"\u003e\n \u003cp\u003e700911\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 200px;\"\u003e\n \u003cp\u003e41.09\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 200px;\"\u003e\n \u003cp\u003eLow risk zone\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 200px;\"\u003e\n \u003cp\u003e335639\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 200px;\"\u003e\n \u003cp\u003e19.68\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e"},{"header":"Discussions","content":"\u003cp\u003eThe HEC-RAS simulated flood and optimized hybrid ANFIS-HEC-RAS model clearly illustrates the flood behaviour of the Olkeriai River basin. HEC-RAS simulated a 100-year return period flood which covered an area of 37.27 km\u0026sup2; with an average depth of 3.22 meters. Small extent coverage from the river corridor is seen on the upper side of the study area with high elevation while large coverage of water is seen on the southern region of the study area due to the low elevation of this region, water pooling being common in such regions causing flooding. Based on these results, immediate measures are therefore needed to curb the dangers of floods in case of any occurrence mostly on the Sothern part of Olkeriai River basin. According to a study by Mcmahon (\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) on recognition and implications of high- depth and energy flood events in Scotland, similar results as these indicate a high-magnitude, high-energy flood event capable of significantly impacting riparian food production in agriculture and proximal settlements. Similar results were reported by Tamiru (\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) on using machine learning and HECRAS integrated models for flood inundation mapping in Baro River Basin, where extreme-event, maximum depth foods were large and perilous to cropland and rural infrastructure. The study indicated that floods in alluvial low-lying floodplains result in extensive flooding and deposition that reduce soil and agricultural yields' quality.\u003c/p\u003e \u003cp\u003eThe spatial water depth and inundation distribution of the HEC-RAS results (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e) are a result of local land use, channel morphological geometry, and local topography. This corroborates with an explanation by Malarchick (\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2019\u003c/span\u003e), describing the hydrogeomorphic and geomorphic response to extreme flood events and how channel confinements and elevation gradient dictate floodwater routing and energies of flow over the duration of extreme events. Though HEC-RAS-enabled useful channel-based predictions of flooding are made simpler, its model is only applied to the defined bank lines ignoring the overbank flow processes or flood extension onto the larger catchment.\u003c/p\u003e \u003cp\u003eThe HEC-RAS-optimized model, extends flood inundation mapping and forecasting to the whole catchment area using spatially distributed data and learning nonlinear relationships from HEC-RAS output, (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e). This was achieved by incorporating multiple floods contributing factors which influence flooding beyond the river corridor. The strong consistency of HEC-RAS and ANFIS-predicted water depths within overlapping areas confirms that the hybrid method effectively surpassed the physical accuracy of the HEC-RAS model while extending its spatial applicability. The optimized output achieved an extremely low RMSE of 0.445 and a virtually zero bias of 0.098, testifying to the strength of the ANFIS-based improvement. These values validate the ability of ANFIS to generalize complex hydrological interactions, particularly under ungauged or data-scarce conditions. This is also evidenced by the work of Bui et al. (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2018\u003c/span\u003e), who used ANFIS to model flood behaviour in the Haraz watershed, Iran and obtained better prediction than that from conventional HEC-HMS model. Likewise, Ullah (\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2013\u003c/span\u003e) highlighted that coupling machine learning with physical models minimizes structural and data-related uncertainties and results in more effective and spatially representative flood simulations.\u003c/p\u003e \u003cp\u003eIn this research, ANFIS enabled flood inundation mapping throughout the entire catchment, beyond river geometry bounded by HEC-RAS, hence facilitating more integrated flood risk assessment. The reclassified results from an optimized model showed regions with different risks of flooding (Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). High risk flooding zones accounted for 39.23% of the total area, occupied most of the area along the river corridor and low elevated zones in the southern part of the area. Moderate risk zones took occupying 41.09% of the catchment area mostly located in low elevated areas in the northern region of the study area while the low-risk zones took the least percentage with only 19.68%. These results call for immediate measures in most of the parts of the study area to mitigate the impending danger of floods in case of occurrence.\u003c/p\u003e \u003cp\u003eResults of this research further attest to the emerging trend of integrating physically based hydraulic models with data-driven models like ANFIS to enhance accuracy and spatial coverage in flood forecasting. To a catchment that is likely to flood with susceptible agriculture and infrastructure, applying these hybrid methods in such a setting has tangible added value in relevant advice for the local planners, policymakers, and emergency managers. Effective maps of floods can direct zoning control, warning infrastructure, and site-based schemes of mitigation like levees or detention ponds to improve community resilience.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis research illustrates that it is possible to achieve much better accuracy and spatial coverage of flood prediction by integrating traditional hydraulic modeling and machine learning algorithms. The novelty is that although HEC-RAS can produce a good simulation of the flood in the river channel, it does not have capacity to simulate the wider floodplain beyond the defined river bank lines. By incorporating ANFIS model which is more data-driven, the shortcomings of the HEC-RAS model were overcome to create a hybrid model with an ability to express the depth as well as the spatial distribution of the flood extremely reliably.\u003c/p\u003e \u003cp\u003eThe optimized model not only improves predictive performance but also allows for more realistic, evidence-based mapping of flood risk throughout the whole catchment. This method allows for more informed decision-making in flood management, land use planning, and disaster preparedness in at-risk river basins. In general, the results support the potential of integrating physically based models with artificial intelligence to develop more comprehensive and beneficial flood risk assessments.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eNo any conflict of interest was declared throughout the research process.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eI wish to acknowledge my supervisors for their support throughout the research process. I also wish to thank the data providers; Water Resource Authority, European Space Agency (ESA), USGS Earth Explorer,\u0026nbsp;Distributed Active Archive Center (DAAC), Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS). Also, I acknowledge the authors I cited in this study for providing vital information for conducting this research work\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe research was funded by the Dutch Research Council (NWO) and the Directorate-General of International Cooperation (DGIS) of the Netherlands Ministry of Foreign Affairs for DUPC3 (2021-2027) Water and Development Partnership Programme of the Smallholder farming families Adapt African Alluvial Aquifers to Strengthen Their Own Resilience (A4Store) Project.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll authors have read, understood, and have complied as applicable with the statement on \u0026ldquo;Ethical responsibilities of Authors\u0026rdquo; as found in the Instructions for Authors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eAccess, O. (2019). \u003cem\u003eThe comparative study of flood modelling with the unsteady and the steady flow on Ngotok river The comparative study of flood modelling with the unsteady and the steady flow on Ngotok river\u003c/em\u003e. https://doi.org/10.1088/1757-899X/669/1/012018\u003c/li\u003e\n \u003cli\u003eAhmed et al. (2007). Summary for policy- makers, climate change, IPCC WG1 fourth assessment report. \u003cem\u003eCambridge Universtity Press\u003c/em\u003e. http://ipcc.ch/meetings/session17/doc3d.pdf%5Cnhttp://link.springer.com/10.1007/BF02986817\u003c/li\u003e\n \u003cli\u003eAl-hmouz, A., Shen, J., Member, S., Al-hmouz, R., \u0026amp; Yan, J. (2012). Modeling and Simulation of an Adaptive Neuro-Fuzzy Inference System ( ANFIS ) for Mobile Learning. \u003cem\u003eIEEE Transactions on Learning Technologies\u003c/em\u003e, \u003cem\u003e5\u003c/em\u003e(3), 226\u0026ndash;237. https://doi.org/10.1109/TLT.2011.36\u003c/li\u003e\n \u003cli\u003eAl, N. et. (2017). Potentials and limitations of 1D, 2D and coupled 1D-2D flood modelling in HEC-RAS. \u003cem\u003eTvvr17/5003\u003c/em\u003e, 8.\u003c/li\u003e\n \u003cli\u003eAshok, V., \u0026amp; Umamahesh, R. N. V. (2019). \u003cem\u003eAssessment of inundation risk in urban floods using HEC RAS 2D\u003c/em\u003e. \u003cem\u003e0123456789\u003c/em\u003e.\u003c/li\u003e\n \u003cli\u003eBernhofen, M. V, Whyman, C., Trigg, M. A., Sleigh, P. A., \u0026amp; Smith, A. M. (2018). \u003cem\u003eA first collective validation of global fluvial flood models for major floods in Nigeria and A fi rst collective validation of global fl uvial fl ood models for major fl oods in Nigeria and Mozambique\u003c/em\u003e.\u003c/li\u003e\n \u003cli\u003eBerthelot, J., Saint-laurent, D., Gervais-beaulac, V., \u0026amp; Savoie, D. (2014). \u003cem\u003eAssessing the Effects of Periodic Flooding on the Population Structure and Recruitment Rates of Riparian Tree Forests\u003c/em\u003e. 2614\u0026ndash;2633. https://doi.org/10.3390/w6092614\u003c/li\u003e\n \u003cli\u003eBui, D. T., Khosravi, K., Li, S., Shahabi, H., Panahi, M., \u0026amp; Ahmad, B. Bin. (2018). \u003cem\u003eNew Hybrids of ANFIS with Several Optimization Algorithms for Flood Susceptibility Modeling\u003c/em\u003e. https://doi.org/10.3390/w10091210\u003c/li\u003e\n \u003cli\u003eCostabile, P., Costanzo, C., Ferraro, D., Macchione, F., \u0026amp; Petaccia, G. (2020). Performances of the new HEC-RAS version 5 for 2-D hydrodynamic-based rainfall-runoff simulations at basin scale: Comparison with a state-of-the art model. \u003cem\u003eWater (Switzerland)\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(9), 1\u0026ndash;19. https://doi.org/10.3390/W12092326\u003c/li\u003e\n \u003cli\u003eCRED. (2015). The Human Cost of Natural Disasters. \u003cem\u003eCentre for Research on the Epidemiology of Disasters CRED\u003c/em\u003e, 58.\u003c/li\u003e\n \u003cli\u003eCunderlik. (2004). Calibration, Verification and Sensitivity analysis of the HEC-HMS hydrologic model. \u003cem\u003eJournal of Chemical Information and Modeling\u003c/em\u003e, \u003cem\u003e53\u003c/em\u003e(9), 1689\u0026ndash;1699.\u003c/li\u003e\n \u003cli\u003eDouben, K. (2006). \u003cem\u003eCHARACTERISTICS OF RIVER FLOODS AND FLOODING : A GLOBAL\u003c/em\u003e. \u003cem\u003e21\u003c/em\u003e(April), 25\u0026ndash;27.\u003c/li\u003e\n \u003cli\u003eDuker, A. E. C., Karimba, B. M., Wani, G. E., \u0026amp; Prasad, P. (2022). \u003cem\u003eSecurity in flexibility : Accessing land and water for irrigation in Kenya-s changing rural environment Security in fl exibility : accessing land and water for irrigation in Kenya \u0026rsquo; s changing rural environment\u003c/em\u003e. \u003cem\u003eMarch\u003c/em\u003e. https://doi.org/10.1051/cagri/2022003\u003c/li\u003e\n \u003cli\u003eEskresi, P. (2015). \u003cem\u003eCoping with Floods\u003c/em\u003e. \u003cem\u003eJuly\u003c/em\u003e. https://doi.org/10.1007/978-94-011-1098-3\u003c/li\u003e\n \u003cli\u003eFeng, B., Zhang, Y., \u0026amp; Bourke, R. (2021). Urbanization impacts on flood risks based on urban growth data and coupled flood models. \u003cem\u003eNatural Hazards\u003c/em\u003e, \u003cem\u003e106\u003c/em\u003e(1), 613\u0026ndash;627. https://doi.org/10.1007/s11069-020-04480-0\u003c/li\u003e\n \u003cli\u003eFernandez. (2009). \u003cem\u003eValidating the Bland-Altman Method of Agreement\u003c/em\u003e. 1\u0026ndash;17.\u003c/li\u003e\n \u003cli\u003eFrench, J., \u0026amp; Ing, R. O. Y. (2010). \u003cem\u003eMortality from Flash Floods : a Review of National Weather Service Reports , 1969-81\u003c/em\u003e. \u003cem\u003eJuly 1981\u003c/em\u003e.\u003c/li\u003e\n \u003cli\u003eGaume, E., Borga, M., Llassat, M. C., Maouche, S., Gaume, E., Borga, M., Llassat, M. C., Maouche, S., \u0026amp; Lang, M. (2016). Mediterranean extreme floods and flash floods To cite this version : \u003cem\u003eThe Mediterranean Region under Climate Change. A Scientific Update\u003c/em\u003e, 133\u0026ndash;144.\u003c/li\u003e\n \u003cli\u003eGuo, K., Guan, M., \u0026amp; Yu, D. (2021). \u003cem\u003eUrban surface water flood modelling \u0026ndash; a comprehensive review of current models and future challenges\u003c/em\u003e. 2843\u0026ndash;2860.\u003c/li\u003e\n \u003cli\u003eHallegatte, S., Ranger, N., Mestre, O., Corfee-morlot, P. D. J., Herweijer, C., \u0026amp; Muir, R. (2022). \u003cem\u003eAssessing climate change impacts , sea level rise and storm surge risk in port cities : a case study on Copenhagen\u003c/em\u003e. 113\u0026ndash;137. https://doi.org/10.1007/s10584-010-9978-3\u003c/li\u003e\n \u003cli\u003eHamerlynck, O., Paul, J., Leauthaud, C., Cochet, H., Nyunja, J., Albergel, J., \u0026amp; Gru, O. (2013). \u003cem\u003eFloods and livelihoods : The impact of changing water resources on wetland agro-ecological production systems in the Tana River Delta , Kenya\u003c/em\u003e. \u003cem\u003e23\u003c/em\u003e, 252\u0026ndash;263.\u003c/li\u003e\n \u003cli\u003eHamill. (2018). Probabilistic precipitation forecast postprocessing using quantile mapping and rank-weighted best-member dressing. \u003cem\u003eMonthly Weather Review\u003c/em\u003e, \u003cem\u003e146\u003c/em\u003e(12), 4079\u0026ndash;4098. https://doi.org/10.1175/MWR-D-18-0147.1\u003c/li\u003e\n \u003cli\u003eHead, R., The, R., Exceedance, A., \u0026amp; Model, P. (2023). \u003cem\u003eRevitalizing the Annual Exceedance Probability model (AEP) for use in the changing world Kaury Pena California State University Maritime Academy\u003c/em\u003e. 1\u0026ndash;35.\u003c/li\u003e\n \u003cli\u003eHouessou-dossou, E. A. Y., Gathenya, J. M., Njuguna, M., \u0026amp; Gariy, Z. A. (2023). \u003cem\u003eFlood Frequency Analysis Using Participatory GIS and Rainfall Data for Two Stations in Narok\u003c/em\u003e. 1\u0026ndash;12.\u003c/li\u003e\n \u003cli\u003eIii. (2005). \u003cem\u003efederal register\u003c/em\u003e.\u003c/li\u003e\n \u003cli\u003eIndexes, V. (2021). \u003cem\u003eEffects of Different Normalization, Aggregation, and Classification Methods on the Construction of Flood Vulnerability Indexes\u003c/em\u003e.\u003c/li\u003e\n \u003cli\u003eJanssens. (2020). Reflection on modern methods: Revisiting the area under the ROC Curve. \u003cem\u003eInternational Journal of Epidemiology\u003c/em\u003e, \u003cem\u003e49\u003c/em\u003e(4), 1397\u0026ndash;1403. https://doi.org/10.1093/ije/dyz274\u003c/li\u003e\n \u003cli\u003eJonkman, S. N., Bočkarjova, M., Kok, M., \u0026amp; Bernardini, P. (2008). Integrated hydrodynamic and economic modelling of flood damage in the Netherlands. \u003cem\u003eEcological Economics\u003c/em\u003e, \u003cem\u003e66\u003c/em\u003e(1), 77\u0026ndash;90. https://doi.org/10.1016/j.ecolecon.2007.12.022\u003c/li\u003e\n \u003cli\u003eKhan, A., Khan, B., Qasim, S., \u0026amp; Khan, S. (2013). Causes, Effects and Remedies: A Case Study of Rural Flooding in District Charsadda, Pakistan. \u003cem\u003eJournal of Managerial \u0026hellip;\u003c/em\u003e.\u003c/li\u003e\n \u003cli\u003eLang\u0026rsquo;at, Kumar, K. (2019). \u003cem\u003eIdentification of the Most Suitable Probability and Mean Streamflow\u003c/em\u003e. 1\u0026ndash;24.\u003c/li\u003e\n \u003cli\u003eMalarchick. (2019). \u003cem\u003eScholarWorks at University of Montana Philosophies of assertiveness Let us know how access to this document benefits you .\u003c/em\u003e\u003c/li\u003e\n \u003cli\u003eMaranzoni, A., Oria, M. D., \u0026amp; Rizzo, C. (2023). \u003cem\u003eQuantitative flood hazard assessment methods : A review\u003c/em\u003e. \u003cem\u003eMarch 2022\u003c/em\u003e, 1\u0026ndash;31. https://doi.org/10.1111/jfr3.12855\u003c/li\u003e\n \u003cli\u003eMasese, A., Neyole, E., \u0026amp; Ombachi, N. (2016). \u003cem\u003eLoss and Damage from Flooding In Lower Nyando Basin , Kisumu County , Kenya\u003c/em\u003e. \u003cem\u003eSeptember\u003c/em\u003e.\u003c/li\u003e\n \u003cli\u003eMatsaba, E. O., van Selm, M., Wesonga, J. M., Goosen, H., Coninx, I., \u0026amp; Koomen, I. (2021). \u003cem\u003eAgricultural Climate Atlas for Kajiado and Kiambu Counties, Kenya BT \u0026nbsp;- Handbook of Climate Change Management: Research, Leadership, Transformation\u003c/em\u003e (J. M. Luetz \u0026amp; D. Ayal (eds.); pp. 3\u0026ndash;23). Springer International Publishing. https://doi.org/10.1007/978-3-030-57281-5_227\u003c/li\u003e\n \u003cli\u003eMcmahon. (2025). \u003cem\u003eHigh-energy flood events recorded in the Mesoproterozoic Meall Dearg Formation , NW Scotland ; their recognition and implications for the study of pre-vegetation alluvium\u003c/em\u003e. \u003cem\u003e175\u003c/em\u003e, 13\u0026ndash;32.\u003c/li\u003e\n \u003cli\u003eMendez, M., Rica, D. C., Rica, C., Jos\u0026eacute;, S., \u0026amp; Rica, C. (2023). \u003cem\u003eA comparison of generalized extreme value , gumbel , and log-pearson distributions for the development of intensity duration frequency curves . A case study in Costa Rica\u003c/em\u003e. \u003cem\u003e01002\u003c/em\u003e, 1\u0026ndash;8.\u003c/li\u003e\n \u003cli\u003eMensah. (2020). Causes , impacts and coping strategies of floods in Ghana : a systematic review. \u003cem\u003eSN Applied Sciences\u003c/em\u003e, \u003cem\u003e2\u003c/em\u003e(5), 1\u0026ndash;13. https://doi.org/10.1007/s42452-020-2548-z\u003c/li\u003e\n \u003cli\u003eMerz, B., Hall, J., Disse, M., \u0026amp; Schumann, A. (2010). Fluvial flood risk management in a changing world. \u003cem\u003eNatural Hazards and Earth System Science\u003c/em\u003e, \u003cem\u003e10\u003c/em\u003e(3), 509\u0026ndash;527. https://doi.org/10.5194/nhess-10-509-2010\u003c/li\u003e\n \u003cli\u003eMfon, I. E., Oguike, M. C., \u0026amp; Eteng, S. U. (2022). \u003cem\u003eCauses and Effects of Flooding in Nigeria : A Review\u003c/em\u003e. \u003cem\u003e1\u003c/em\u003e(9), 1777\u0026ndash;1792.\u003c/li\u003e\n \u003cli\u003eMorris. (2014). \u003cem\u003eThe impact of seasonal flooding on agriculture : the spring 2012 floods in Somerset , England\u003c/em\u003e. \u003cem\u003e7\u003c/em\u003e(April 2012), 128\u0026ndash;140. https://doi.org/10.1111/jfr3.12041\u003c/li\u003e\n \u003cli\u003eMunyai, R. B., Chikoore, H., Musyoki, A., Chakwizira, J., Muofhe, T. P., Xulu, N. G., \u0026amp; Manyanya, T. C. (2021). \u003cem\u003eVulnerability and Adaptation to Flood Hazards in Rural Settlements of Limpopo Province , South Africa\u003c/em\u003e.\u003c/li\u003e\n \u003cli\u003eNjogu, H. W. (2021). Effects of floods on infrastructure users in Kenya. \u003cem\u003eJournal of Flood Risk Management\u003c/em\u003e, \u003cem\u003e14\u003c/em\u003e(4), 1\u0026ndash;10. https://doi.org/10.1111/jfr3.12746\u003c/li\u003e\n \u003cli\u003eNordlo. (2016). A review on mechanism of flood disaster management in Asia. \u003cem\u003eInternational Review of Management and Marketing\u003c/em\u003e, \u003cem\u003e6\u003c/em\u003e(1), 29\u0026ndash;52.\u003c/li\u003e\n \u003cli\u003eNow, A. (2018). \u003cem\u003eGlobal Flooding\u003c/em\u003e. \u003cem\u003e8200\u003c/em\u003e(2017), 47\u0026ndash;52. https://doi.org/10.1080/19428200.2017.1390365\u003c/li\u003e\n \u003cli\u003eOgega, O. M. (2018). Globalization and Global Warming: A Case of Laikipia County, Kenya. \u003cem\u003eInternational Conference on Climate Change \u0026amp; Sustainable Development in Africa 2018\u003c/em\u003e, \u003cem\u003e1\u003c/em\u003e(Special Issue: Environment, Technology and Sustainable Development), 60\u0026ndash;67. https://doi.org/10.26796/jenrm.v0i0.106\u003c/li\u003e\n \u003cli\u003eOkaka. (2019). Health vulnerability to flood-induced risks of households in flood-prone informal settlements in the Coastal City of Mombasa, Kenya. \u003cem\u003eNatural Hazards\u003c/em\u003e, \u003cem\u003e99\u003c/em\u003e(2), 1007\u0026ndash;1029. https://doi.org/10.1007/s11069-019-03792-0\u003c/li\u003e\n \u003cli\u003ePalakonda, V., \u0026amp; Mallipeddi, R. (2017). Pareto Dominance-Based Algorithms With Ranking Methods for Many-Objective Optimization. \u003cem\u003eIEEE Access\u003c/em\u003e, \u003cem\u003e5\u003c/em\u003e, 11043\u0026ndash;11053. https://doi.org/10.1109/ACCESS.2017.2716779\u003c/li\u003e\n \u003cli\u003ePiracha. (2021). \u003cem\u003eNatural Disasters \u0026mdash; Origins , Impacts , Management\u003c/em\u003e. 1101\u0026ndash;1131.\u003c/li\u003e\n \u003cli\u003eRannie, W. (2016). The 1997 flood event in the Red River basin: Causes, assessment and damages. \u003cem\u003eCanadian Water Resources Journal\u003c/em\u003e, \u003cem\u003e41\u003c/em\u003e(1\u0026ndash;2), 45\u0026ndash;55. https://doi.org/10.1080/07011784.2015.1004198\u003c/li\u003e\n \u003cli\u003eRisa. (2016). \u003cem\u003eFlooding as an environmental hazard\u003c/em\u003e.\u003c/li\u003e\n \u003cli\u003eRisi, R. De, Jalayer, F., Paola, F. De, \u0026amp; Lindley, S. (2018). Delineation of flooding risk hotspots based on digital elevation model , calculated and historical flooding extents : the case of Ouagadougou. \u003cem\u003eStochastic Environmental Research and Risk Assessment\u003c/em\u003e, \u003cem\u003e32\u003c/em\u003e(6), 1545\u0026ndash;1559. https://doi.org/10.1007/s00477-017-1450-8\u003c/li\u003e\n \u003cli\u003eSaudi. (2023). \u003cem\u003eMapping of Groundwater, Flood, and Drought Potential Zones in Neom, Saudi Arabia, Using GIS and Remote Sensing Techniques\u003c/em\u003e.\u003c/li\u003e\n \u003cli\u003eSetiawan, M. N., Kristanto, D., \u0026amp; Setiawan, J. (2018). \u003cem\u003eEstimating design flood and HEC-RAS modelling approach for flood analysis in Bojonegoro city Estimating design flood and HEC-RAS modelling approach for flood analysis in Bojonegoro city\u003c/em\u003e. https://doi.org/10.1088/1757-899X/316/1/012042\u003c/li\u003e\n \u003cli\u003eShabir. (2016). One Dimensional Steady Flow Analysis Using HECRAS \u0026ndash; A case of River Jhelum, Jammu and Kashmir. \u003cem\u003eEuropean Scientific Journal, ESJ\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(32), 340. https://doi.org/10.19044/esj.2016.v12n32p340\u003c/li\u003e\n \u003cli\u003eTamiru. (2022). Machine ‑ learning and HEC ‑ RAS integrated models for flood inundation mapping in Baro River Basin , Ethiopia. \u003cem\u003eModeling Earth Systems and Environment\u003c/em\u003e, \u003cem\u003e8\u003c/em\u003e(2), 2291\u0026ndash;2303. https://doi.org/10.1007/s40808-021-01175-8\u003c/li\u003e\n \u003cli\u003eTramblay, Y., Villarini, G., \u0026amp; Zhang, W. (2020). Observed changes in flood hazard in Africa. \u003cem\u003eEnvironmental Research Letters\u003c/em\u003e, \u003cem\u003e15\u003c/em\u003e(10). https://doi.org/10.1088/1748-9326/abb90b\u003c/li\u003e\n \u003cli\u003eTshimanga, R. M., Marie, T. J., Kabuya, P., \u0026amp; Alsdorf, D. (2016). \u003cem\u003eFlood forecasting : an international perspective A Regional Perceptive of Flood Forecasting and Disaster Management Systems for the Congo River Basin\u003c/em\u003e. \u003cem\u003eApril 2017\u003c/em\u003e.\u003c/li\u003e\n \u003cli\u003eUker, A. E. C., Karimba, B. M., Wani, G. E., Prasad, P., Zaag, P. Vd. Der, \u0026amp; Fraiture, C. De. (2022). \u003cem\u003eSecurity in fl exibility : accessing land and water for irrigation in Kenya \u0026rsquo; s changing rural environment\u003c/em\u003e.\u003c/li\u003e\n \u003cli\u003eUllah, N. (2013). \u003cem\u003eFlood Flow Modeling in a River System Using Adaptive Neuro-Fuzzy Inference System\u003c/em\u003e. \u003cem\u003e2\u003c/em\u003e(2), 54\u0026ndash;68. https://doi.org/10.5296/emsd.v2i2.3738\u003c/li\u003e\n \u003cli\u003eUSACE. (1998). HEC-HMS Technical Reference Manual Introduction. \u003cem\u003eComputer Manual\u003c/em\u003e, 1\u0026ndash;288.\u003c/li\u003e\n \u003cli\u003eWahyudi. (2020). The analysis of the causes of flood disasters and their impacts in the perspective of environmental law. \u003cem\u003eIOP Conference Series: Earth and Environmental Science\u003c/em\u003e, \u003cem\u003e437\u003c/em\u003e(1). https://doi.org/10.1088/1755-1315/437/1/012056\u003c/li\u003e\n \u003cli\u003eWalia, N. (2015). \u003cem\u003eANFIS : Adaptive Neuro-Fuzzy Inference System- A Survey\u003c/em\u003e. \u003cem\u003e123\u003c/em\u003e(13), 32\u0026ndash;38.\u003c/li\u003e\n \u003cli\u003eWorld. (2024). \u003cem\u003eGlobal , regional and national trends and impacts of natural floods ,\u003c/em\u003e. \u003cem\u003e38\u003c/em\u003e, 410\u0026ndash;420.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"modeling-earth-systems-and-environment","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"mese","sideBox":"Learn more about [Modeling Earth Systems and Environment](http://link.springer.com/journal/40808)","snPcode":"40808","submissionUrl":"https://submission.springernature.com/new-submission/40808/3","title":"Modeling Earth Systems and Environment","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Flood modelling, flood resilience, HEC-RAS, ANFIS, HEC-HMS, hybrid model, flood hazard map","lastPublishedDoi":"10.21203/rs.3.rs-6670542/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6670542/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eEffective flood modelling is essential in flood disasters\u0026rsquo; impact reduction and sustainable land use planning, particularly in vulnerable areas such as the Olkeriai River Basin in Kenya. This study provides an innovative hybrid model of Adaptive Neuro-Fuzzy Inference System (ANFIS) and the Hydrologic Engineering Center-River Analysis System (HEC-RAS) model for improved spatial accuracy in flood inundation mapping. The coupled model provides flood inundation mapping for the whole catchment area unlike HEC-RAS model which is restricted to the defined river bank lines. Flooding in the Olkeriai River basin continues to disrupt riparian agriculture and settlements in this basin, but most conventional hydrological models tend to not accurately simulate flood extents over varied terrain. The steady flow simulation in HEC-RAS was used to simulate a 100-yr return period flood with peak flows from a calibrated hydrologic model in Hydrologic Engineering Center- Hydrologic Modelling System (HEC-HMS). Historical events of flooding and conditioning factors were used to train ANFIS model to create a spatial flood Inundation index map. Lastly, HEC-RAS flood depth inundation outputs were calibrated by overlaying them on the flood inundation index map based on ANFIS model outputs. Results indicate that ANFIS model worked well in terms of accuracy and prediction (R\u0026sup2; = 0.960, RMSE\u0026thinsp;=\u0026thinsp;0.092, MAE\u0026thinsp;=\u0026thinsp;0.090 and AOC\u0026thinsp;=\u0026thinsp;0.910), and hybrid model enhanced flood prediction capability (R\u0026sup2;= 0.944, RMSE\u0026thinsp;=\u0026thinsp;0.445, MAE\u0026thinsp;=\u0026thinsp;0.337 and NSE\u0026thinsp;=\u0026thinsp;0.944). Derived flood inundation map delineates the high-risk areas within and outside the river corridor. These outcomes will enable local authorities, disaster managers, and planners to implement effective actions in flood mitigation, plan early warnings, and assist land-use planning that renders the community more resilient.\u003c/p\u003e","manuscriptTitle":"Optimization of the HEC-RAS Based Flood Inundation Mapping using Adaptive Neuro- Fuzzy Inference System: A case study of Olkeriai River Basin, Kenya","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-05-29 12:20:29","doi":"10.21203/rs.3.rs-6670542/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-09-29T19:27:02+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-07T19:39:02+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-06-20T09:01:21+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"250108665221655989975338600228317972868","date":"2025-05-31T04:36:14+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"243698336632115508310446676552344639357","date":"2025-05-29T08:43:17+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"202911290747789708444547927766727763843","date":"2025-05-28T18:06:40+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"303848411086018027849916491148835739367","date":"2025-05-28T16:45:45+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"78805128722789213423443161585476298863","date":"2025-05-28T14:35:48+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-05-28T14:18:15+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-05-26T07:21:14+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-05-26T07:16:52+00:00","index":"","fulltext":""},{"type":"submitted","content":"Modeling Earth Systems and Environment","date":"2025-05-15T08:24:23+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"modeling-earth-systems-and-environment","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"mese","sideBox":"Learn more about [Modeling Earth Systems and Environment](http://link.springer.com/journal/40808)","snPcode":"40808","submissionUrl":"https://submission.springernature.com/new-submission/40808/3","title":"Modeling Earth Systems and Environment","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"956a1f00-9363-4d49-be13-a52ca8518cd2","owner":[],"postedDate":"May 29th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-11-10T16:00:26+00:00","versionOfRecord":{"articleIdentity":"rs-6670542","link":"https://doi.org/10.1007/s40808-025-02667-7","journal":{"identity":"modeling-earth-systems-and-environment","isVorOnly":false,"title":"Modeling Earth Systems and Environment"},"publishedOn":"2025-11-05 15:57:24","publishedOnDateReadable":"November 5th, 2025"},"versionCreatedAt":"2025-05-29 12:20:29","video":"","vorDoi":"10.1007/s40808-025-02667-7","vorDoiUrl":"https://doi.org/10.1007/s40808-025-02667-7","workflowStages":[]},"version":"v1","identity":"rs-6670542","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6670542","identity":"rs-6670542","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}